Compare commits

...

44 Commits

Author SHA1 Message Date
Dotta
bab5136645 Keep handoff resolution logging non-fatal 2026-05-05 12:16:30 -05:00
Dotta
09ed4e54cb Add operator QoL screenshots 2026-05-05 12:06:21 -05:00
Dotta
783f4d2f28 Address operator QoL PR feedback 2026-05-05 11:59:50 -05:00
Dotta
433326ffcb Improve operator workflow QoL
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-05-05 11:34:01 -05:00
Dotta
3c73ed26b5 Expand plugin host surface (#5205)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - The plugin system is the extension boundary for optional product
capabilities
> - Rich plugins need more than a worker entrypoint: they need scoped
database storage, local project folders, managed agents/routines, host
navigation, and reusable UI components
> - The LLM Wiki work exposed those missing host surfaces while keeping
plugin code outside the core control plane
> - This pull request expands the core plugin host, SDK, server APIs,
and UI bridge so plugins can declare and use those surfaces
> - The benefit is that future plugins can integrate with Paperclip
through documented, validated contracts instead of bespoke server or UI
imports

## What Changed

- Added plugin-managed database namespaces and migration tracking,
including Drizzle schema/migration files and SQL validation for
namespace isolation.
- Added server support for plugin local folders, managed agents, managed
routines, scoped plugin APIs, and plugin operation visibility.
- Expanded shared plugin manifest/types/validators and SDK
host/testing/UI exports for richer plugin surfaces.
- Added reusable UI pieces for file trees, managed routines, resizable
sidebars, route sidebars, and plugin bridge initialization.
- Updated plugin docs and example plugins to use the expanded host and
SDK surface.

## Verification

- `pnpm install --frozen-lockfile`
- `pnpm run preflight:workspace-links && pnpm exec vitest run
packages/shared/src/validators/plugin.test.ts
server/src/__tests__/plugin-database.test.ts
server/src/__tests__/plugin-local-folders.test.ts
server/src/__tests__/plugin-managed-agents.test.ts
server/src/__tests__/plugin-managed-routines.test.ts
server/src/__tests__/plugin-orchestration-apis.test.ts
ui/src/api/plugins.test.ts ui/src/components/FileTree.test.tsx
ui/src/components/ResizableSidebarPane.test.tsx
ui/src/pages/PluginPage.test.tsx ui/src/plugins/bridge.test.ts` passed:
11 files, 67 tests.
- Confirmed this PR changes 89 files and does not include
`pnpm-lock.yaml` or `.github/workflows/*`.

## Risks

- Medium: this expands plugin host contracts across db/shared/server/ui
and includes a new core migration (`0076_useful_elektra.sql`).
- The plugin database namespace validator is intentionally restrictive;
plugin authors may need follow-up affordances for SQL patterns that
remain blocked.
- Merge this before the LLM Wiki plugin PR so the plugin can resolve the
new SDK and host APIs.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool-enabled shell/git/GitHub
workflow. Context window size was not exposed by the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-05 07:42:57 -05:00
Dotta
d6bee62f02 Fix Cloud tenant issue identifier routes (#5196)
## Summary

- Allow Cloud tenant issue identifiers with alphanumeric prefixes, such
as `PC1897-1`, to normalize as issue references.
- Resolve those identifiers through issue detail/update routes, active
run/live run polling, activity, costs, and `issueService.getById`.
- Keep UI issue-link parsing aligned so tenant links normalize back to
`/issues/<IDENTIFIER>`.

## Root Cause

Cloud tenant issue prefixes include digits from the stack-id hash. The
app-side route normalization still accepted only all-letter prefixes, so
`/api/issues/PC1897-1` skipped identifier lookup and fell through as a
non-UUID id.

## Verification

- `pnpm exec vitest run packages/shared/src/issue-references.test.ts
ui/src/lib/issue-reference.test.ts
server/src/__tests__/issue-identifier-routes.test.ts
server/src/__tests__/activity-routes.test.ts
server/src/__tests__/costs-service.test.ts
server/src/__tests__/agent-live-run-routes.test.ts
server/src/__tests__/issues-service.test.ts`
- `pnpm --filter @paperclipai/shared typecheck && pnpm --filter
@paperclipai/server typecheck`
- `git diff --check`

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-04 13:20:58 -05:00
Dotta
edbb670c3b Merge pull request #5154 from paperclipai/pap-3474-docker-timeout
Raise Docker image build timeout
2026-05-03 23:01:46 -05:00
Dotta
fd10404374 Raise Docker image build timeout
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-05-03 22:52:33 -05:00
Devin Foley
47920f9c47 Speed up PR CI critical path (#5147)
## Thinking Path

> - Paperclip orchestrates AI agents for autonomous companies, so
developer throughput on the control plane repo directly affects how fast
the product can evolve.
> - The PR workflow is part of that throughput surface because every
change waits on it before review and merge.
> - This branch started from measured evidence that the PR critical path
was dominated by work that was either serialized unnecessarily or placed
on the wrong part of the graph.
> - The biggest concrete problems were: the canary dry run living inside
`verify`, the server isolated suites running one-by-one in a single
lane, and duplicate CI work that the PR path was paying for without
increasing coverage proportionally.
> - This pull request restructures the PR workflow so those costs are
reduced without removing the important coverage that was already
protecting release and test quality.
> - Follow-up fixes on the branch hardened the new entrypoints so they
work on clean GitHub runners and so the reduced PR typecheck path stays
self-maintaining as workspace packages evolve.
> - The benefit is materially faster PR wall-clock time while keeping
canary packaging checks, serialized-suite isolation, plugin SDK
consumers, and explicit TypeScript coverage where builds do not already
provide it.

## What Changed

- Moved the PR canary dry run into its own `Canary Dry Run` job so it
still runs on PRs but no longer extends the `verify` critical path.
- Split the custom Vitest runner into `general`, `serialized`, and `all`
modes, and added shard support for the isolated server suites.
- Added `test:run:general` and `test:run:serialized` scripts, then
rewired PR CI to fan the serialized server suites out across a 4-way
matrix.
- Added the required `@paperclipai/plugin-sdk` build preflight before
the new reduced-scope typecheck and test entrypoints so they succeed on
clean CI runners.
- Replaced the hardcoded PR build-gap list with
`scripts/run-typecheck-build-gaps.mjs`, which discovers workspace
packages whose `build` scripts skip TypeScript and runs only their
explicit `typecheck` scripts.
- Removed the redundant `pnpm build` from the PR `e2e` job because the
Playwright onboarding path boots Paperclip from source.

## Verification

- `ruby -e "require 'yaml'; YAML.load_file('.github/workflows/pr.yml');
puts 'workflow ok'"`
- `node scripts/run-vitest-stable.mjs --mode general --dry-run`
- `node scripts/run-vitest-stable.mjs --mode serialized --shard-index 0
--shard-count 4 --dry-run`
- `pnpm run typecheck:build-gaps`
- `pnpm test:run:general`
- `pnpm test:run:serialized -- --shard-index 0 --shard-count 4`
- `pnpm build`
- `pnpm paperclipai onboard --yes --run`
- `curl http://127.0.0.1:3299/api/health`

## Risks

- Branch protection or required-check configuration may need to be
updated for the new standalone `Canary Dry Run` job and the
serialized-suite matrix job names.
- `scripts/run-typecheck-build-gaps.mjs` assumes packages that need
explicit PR-time typechecking are the ones whose `build` scripts omit
`tsc`; if build conventions change, that heuristic needs to stay
aligned.
- Serialized test sharding preserves per-suite isolation, but the first
few CI runs should still be watched for shard-balance or naming
assumptions in downstream tooling.

## Model Used

- OpenAI GPT-5.4 via the Codex local adapter, using high reasoning
effort with shell, git, and file-edit tool use in a local worktree.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-03 20:20:14 -07:00
Dotta
e01ffc18d3 Merge pull request #5148 from paperclipai/pap-3474-tenant-identity-deploy
Support Cloud tenant identity bootstrap
2026-05-03 21:57:28 -05:00
Dotta
ae23e02526 Support Cloud tenant identity bootstrap
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-05-03 21:55:52 -05:00
Devin Foley
29401b231b fix(ci): gate new release packages on npm bootstrap (#5146)
## Thinking Path

> - Paperclip is a control plane for autonomous agent companies, so its
release automation is part of the core operator trust boundary.
> - The affected subsystem is npm/GitHub Actions release publishing for
the public monorepo packages.
> - The concrete failure was that a newly added package reached
`master`, the canary workflow attempted its first publish, and npm
trusted publishing was not yet bootstrapped for that package.
> - That means the problem is not just one broken run; it is a missing
pre-merge guard that lets release-ineligible packages land and only fail
once `publish_canary` runs.
> - This pull request makes release enrollment explicit, validates that
enrollment in CI, and adds a PR-time bootstrap check against npm for
changed release-enabled package manifests.
> - The result is that we keep trusted publishing, avoid teaching CI to
`npm adduser`, and move this class of failure from post-merge canary
time to pre-merge review time.

## What Changed

- Added `scripts/release-package-manifest.json` so release-managed
public packages are explicitly enrolled instead of being inferred from
every non-private workspace package.
- Hardened `scripts/release-package-map.mjs` to validate the manifest
before release workflows rewrite versions or assemble publish payloads.
- Added `scripts/check-release-package-bootstrap.mjs` and wired it into
`.github/workflows/pr.yml` so PRs that change a release-enabled package
manifest fail if that package does not already exist on npm.
- Added release-package manifest coverage tests to
`scripts/release-package-map.test.mjs` and included them in `pnpm run
test:release-registry`.
- Wired manifest validation into `.github/workflows/release.yml` and
documented the first-publish bootstrap policy in `doc/PUBLISHING.md` and
`doc/RELEASE-AUTOMATION-SETUP.md`.

## Verification

- `pnpm run test:release-registry`
- `./scripts/release.sh canary --skip-verify --dry-run`
- Confirmed the committed diff contains no obvious PII/secrets via
targeted pattern scan before pushing.

## Risks

- Low risk overall: this is CI/release-policy code, not product runtime
logic.
- The new PR bootstrap check depends on npm metadata availability, so a
transient npm outage could block a PR that changes a release-enabled
package manifest.
- The manifest introduces a new source of truth that must stay aligned
with public package additions, but that is intentional and now enforced.

## Model Used

- OpenAI Codex via the `codex_local` Paperclip adapter; GPT-5-based
coding agent with tool use, terminal execution, git, and GitHub CLI.
Exact served model ID/context window are not exposed by the local
runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 19:31:28 -07:00
Devin Foley
a5430f010d Handle Gemini assistant message events in JSONL parser (#5143)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies, including
agents
>   running the Gemini CLI (`gemini-local` adapter)
> - The Gemini CLI emits a JSONL event stream during a run that the
adapter
> parses to extract the assistant's response text, tool results, and
usage
> - Recent versions of the Gemini CLI emit assistant responses as
> `{ "type": "message", "role": "assistant", "content": ... }` events in
>   addition to the previously-handled event shapes
> - The parser was not handling the new event type, so the assistant's
actual
> response text was being silently dropped from parsed output. Callers
ended
>   up with empty assistant messages even when Gemini had successfully
>   responded
> - This PR teaches the parser to recognize `{type: "message", role:
>   "assistant"}` events and extract their content text via the same
>   `collectMessageText` helper used for other message-shaped events
> - The benefit is that Gemini runs surface the assistant's real
response in
> downstream consumers (issue comments, run logs, downstream agent
context)
>   instead of vanishing

## What Changed

- `packages/adapters/gemini-local/src/server/parse.ts`: in
`parseGeminiJsonl(...)`, add a branch for `event.type === "message"`
with
  `role === "assistant"` that calls
  `messages.push(...collectMessageText(event.content))`.
- `packages/adapters/gemini-local/src/server/parse.test.ts`: ~19 lines
of
  coverage for the new branch.

## Verification

- `pnpm --filter @paperclipai/adapter-gemini-local test -- parse`
- Manual QA: run a Gemini agent on an issue, confirm the assistant's
response
appears as the issue comment / run output. Before this fix the comment
was
  empty even when the run completed successfully.

## Risks

- Tightly scoped: 8 lines of production code in one parser branch. No
effect
  on existing event shapes or other adapters.
- If the Gemini CLI changes its event schema again, this branch may need
to be
  revisited — but adding it is strictly additive over current behaviour.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 18:36:50 -07:00
Devin Foley
6c090f84a9 Strip inherited host shell env from SSH remote execution (#5142)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agents executing on remote SSH hosts receive an env map built from
the host
>   process's env plus per-run additions like `PAPERCLIP_API_KEY`,
>   `PAPERCLIP_RUN_ID`, etc.
> - The env map currently includes inherited host vars by default,
including
> identity-bound ones like `PATH`, `HOME`, `USER`, `NVM_DIR`, `XDG_*` —
> variables whose values are meaningful only on the host they came from
> - Sending the host's `PATH` (containing host-only directories like a
local
> nvm install path) to a remote SSH box overrides the remote's actual
`PATH`
> and breaks command resolution. Same hazard for `HOME` (commands
looking for
> config files end up in a non-existent dir), `USER` (writes go to the
wrong
>   path), etc.
> - This PR adds `sanitizeSshRemoteEnv()` that drops inherited
identity-bound
> vars when their value matches the host process's value. Explicitly-set
> values pass through untouched, so callers that genuinely want to
override
>   remote `PATH` etc. still can — but accidental leakage from
>   `process.env` is filtered.
> - The benefit is that SSH remote execution stops corrupting the remote
> shell's environment with host-shaped paths, so commands resolve
correctly
>   against the remote PATH and config files land in the remote `HOME`

## What Changed

- New `sanitizeSshRemoteEnv(env, inheritedEnv = process.env)` in
`packages/adapter-utils/src/server-utils.ts`. The identity-bound key set
is:
    - `PATH`, `HOME`, `PWD`, `SHELL`, `USER`, `LOGNAME`
    - `NVM_DIR`, `TMPDIR`, `TMP`, `TEMP`
    - `XDG_CONFIG_HOME`, `XDG_CACHE_HOME`, `XDG_DATA_HOME`,
      `XDG_STATE_HOME`, `XDG_RUNTIME_DIR`
For any key in this set, the entry is dropped iff the env value equals
the
  inherited (host process) value. Other keys pass through unchanged.
- `readEnvValueCaseInsensitive(...)` helper handles Windows-style
  case-insensitive env var lookups.
- Wired into `resolveSpawnTarget(...)` for the SSH transport. Sandbox
and local
  paths are unaffected.
- Tests added in `server-utils.test.ts` (~50 lines) covering: matching
keys
filtered, mismatched keys preserved, non-identity keys passed through,
case
  insensitivity.

## Verification

- `pnpm --filter @paperclipai/adapter-utils test -- server-utils`
- Manual QA: run any adapter against an SSH-backed environment, confirm
remote command resolution works (e.g. `node`, `npm`, the adapter's CLI)
and
config files land in the remote user's `HOME`. Compare to the prior
behaviour
by transiently re-introducing the inherited `PATH` and watching commands
  fail with `command not found`.

## Risks

- Behavioural shift: SSH remote execution previously passed inherited
host env
  vars verbatim. Code that relied on that (e.g. a remote command somehow
  expecting the host's `PATH`) will see different behaviour. None of the
  adapter code in this repo has such a dependency.
- Edge case: if a caller explicitly sets `PATH` to the same value as the
host's
`PATH` (literally — same exact string), the sanitizer drops it as a
leak.
  In practice no caller constructs the env this way.
- Windows host: case-insensitive lookup handles `Path` vs `PATH`
correctly.
  Tested.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 18:36:13 -07:00
Devin Foley
90631b09b3 Let adapters declare runtime command spec for remote provisioning (#5141)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies, running
adapter
> commands like `claude`, `codex`, `pi` either locally or on remote
runtimes
>   (SSH hosts, sandboxes, etc.)
> - On a fresh remote runtime — particularly an ephemeral sandbox — the
> adapter's CLI may not be installed yet. Today operators handle this
via
> external configuration (e.g. a project-level `provisionCommand` shell
> script) that has to know about every adapter the operator might want
to use
> - This means every adapter has its own well-known npm package, but
operators
>   end up writing duplicate provision shell scripts that paste together
> `npm install -g @anthropic-ai/claude-code`, `npm install -g
@openai/codex`,
>   etc. — knowledge the adapter itself already has
> - This PR moves that knowledge into the adapter modules: each adapter
declares
> how its runtime command should be detected and (if applicable)
installed
> via `getRuntimeCommandSpec(config)`. The execution path runs the
adapter's
> own install command on remote sandbox targets before launching, so a
fresh
> sandbox bootstraps itself instead of requiring a hand-written
provision script
> - The benefit is fewer footguns for operators provisioning remote
runtimes,
>   and a clean place for new adapters to plug in their install recipe

## What Changed

- New types in `packages/adapter-utils/src/types.ts`:
    - `AdapterRuntimeCommandSpec` describing `command`, optional
      `detectCommand`, and optional `installCommand`
    - Optional `getRuntimeCommandSpec(config)` on `ServerAdapterModule`
- Optional `runtimeCommandSpec` on `AdapterExecutionContext` so adapters
      receive the resolved spec at execute time
- New helper `ensureAdapterExecutionTargetRuntimeCommandInstalled(...)`
in
`packages/adapter-utils/src/execution-target.ts` that runs the install
command
on remote targets when `transport === "sandbox"`. SSH and local targets
are
  no-ops. Throws on timeout or non-zero exit so failures surface early.
- Each of `claude-local`, `codex-local`, `cursor-local`, `gemini-local`,
  `opencode-local`, `pi-local`'s `execute.ts` now reads
`ctx.runtimeCommandSpec?.installCommand` and calls the helper before
launching
  the adapter command.
- `server/src/adapters/registry.ts` declares `getRuntimeCommandSpec` for
each
  adapter:
- claude/codex/gemini/opencode/pi-local: `npm install -g <package>`
recipe via
a shared `buildNpmRuntimeCommandSpec` helper, with a defensive guard
that
only auto-installs when the configured `command` matches the well-known
      fallback (custom binaries are left alone).
- cursor-local: declares `command` only; no auto-install (no public npm
      package), preserving the existing manual setup.
- `server/src/services/heartbeat.ts` resolves the spec via
`adapter.getRuntimeCommandSpec?.(runtimeConfig)` and passes it through
to
  `AdapterExecutionContext`.
- Tests added in `execution-target.test.ts` (~75 lines), e2b
`plugin.test.ts` (~32 lines), and `environment-run-orchestrator.test.ts`
  (~76 lines).

## Verification

- `pnpm --filter @paperclipai/adapter-utils test`
- `pnpm --filter @paperclipai/server test --
environment-run-orchestrator`
- `pnpm --filter @paperclipai/sandbox-providers-e2b test`
- Manual QA: run an adapter (claude/codex/etc.) against a fresh
sandbox-backed
environment that does NOT have the adapter CLI pre-installed. Confirm
the
install runs once at the start of the agent run and the adapter then
launches
successfully. Re-run on the same sandbox; confirm the install command is
  idempotent and the second run starts faster.
- Confirm SSH and local execution paths are unaffected (gated by
  `transport === "sandbox"`).

## Risks

- Behavioural shift on sandbox runs: a new install step now runs at the
start
  of every sandbox agent run for adapters with `installCommand` set. The
install commands are idempotent (`if ! command -v X >/dev/null 2>&1;
then
npm install -g <pkg>; fi`), so this is fast on warm sandboxes. On a cold
  sandbox, the first run takes longer.
- Operators who used the legacy project-level `provisionCommand` to
install
adapter CLIs can drop that part of their script; the adapter handles it
now.
  Existing scripts continue to work — installs are idempotent.
- The cursor-local adapter has no auto-install (no public npm package).
  Behaviour for cursor-local on sandboxes is unchanged.
- New optional surface on `ServerAdapterModule`. Plugins that don't
implement
  `getRuntimeCommandSpec` retain previous behaviour (no auto-install).

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 18:35:36 -07:00
Devin Foley
2dce81fbf6 Add optional bridge proxy request logging via PAPERCLIP_BRIDGE_DEBUG (#5140)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agents on remote sandboxes call back into the Paperclip control
plane via a
> callback bridge — the host process running the bridge proxies HTTP
requests
>   from the sandbox to the Paperclip API
> - When something goes wrong end-to-end (sandbox can't reach Paperclip,
requests
> timing out, malformed responses), it's hard to tell whether the bridge
> processed the request, what URL/method it used, and what the upstream
>   responded with
> - There was no built-in way to log bridge proxy traffic without
modifying
>   adapter code or attaching a debugger
> - This PR adds opt-in stdout logging of every bridge proxy request and
response
> (method, path, query, status), gated behind `PAPERCLIP_BRIDGE_DEBUG`
so it
>   stays off by default
> - The benefit is that operators can flip a single env var to get full
visibility
> into bridge traffic when debugging remote runs, without changing code

## What Changed

- `packages/adapter-utils/src/execution-target.ts`:
`startAdapterExecutionTargetPaperclipBridge`'s `handleRequest` now logs
each
  proxied request and response when `PAPERCLIP_BRIDGE_DEBUG` is truthy:
    - `[paperclip] Bridge proxy <METHOD> <path>?<query>` before fetch
- `[paperclip] Bridge proxy response <status> for <METHOD>
<path>?<query>` after
- Logging is no-op when the env var is unset/`"0"`/`"false"`.

## Verification

- Set `PAPERCLIP_BRIDGE_DEBUG=1` in the host process env, run an agent
against
a sandbox-backed environment, confirm the bridge log lines appear in
stdout.
- Unset the env var and confirm no extra log lines appear during normal
runs.

## Risks

- Off-by-default, no observable change for shipping users.
- When enabled, the logging is verbose — every API call from the sandbox
  produces 2 stdout lines. Operators should only enable it during active
  debugging.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [ ] I have added or updated tests where applicable — covered by
exercising the
      flag in dev; the underlying handleRequest behavior is unchanged
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 18:34:48 -07:00
Devin Foley
0e51fa2b0d Honor reuse-existing preference and assignee default environment in issue runs (#5139)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agents run inside execution workspaces (a per-issue cwd + env), and
an issue
> can prefer to reuse an existing workspace or get a fresh one each time
> - The heartbeat service was reading the existing workspace's config to
derive
> environment selection regardless of whether the issue actually wanted
to reuse
> it. So fresh-run issues were inheriting stale config from a workspace
that was
>   about to be discarded
> - Separately, when an issue is assigned to an agent, the issue's
execution
> workspace settings weren't picking up the agent's
`defaultEnvironmentId`,
>   even though the agent's choice is the natural default for that issue
> - This PR makes both selection paths honor the obvious source of
truth:
> workspace config flows only when the issue actually wants
`reuse_existing`,
> and the assignee agent's default environment is applied at assignment
time if
>   nothing else is set on the issue
> - The benefit is that re-running a flaky issue picks up the right
environment
> instead of inheriting the previous run's config, and assigning an
agent to an
>   issue does the obvious thing without operator intervention

## What Changed

- `server/src/services/heartbeat.ts`: introduce
`reusableExecutionWorkspaceConfig`
  that is non-null only when `shouldReuseExisting` is true. Both
  `resolveExecutionWorkspaceEnvironmentId(...)` and
`applyPersistedExecutionWorkspaceConfig(...)` now read from it instead
of
unconditionally consulting `existingExecutionWorkspace?.config`.
Fresh-run
issues no longer inherit stale environment config from an in-flight
workspace
  about to be discarded.
- `server/src/services/issues.ts`: when an issue update sets a new
  `assigneeAgentId` and isolated workspaces are enabled, populate
  `executionWorkspaceSettings.environmentId` from the assignee agent's
  `defaultEnvironmentId` if the issue doesn't have an explicit
  `environmentId` set yet.
- Tests added in `heartbeat-plugin-environment.test.ts` (~216 lines) and
  `issues-service.test.ts` (~85 lines) covering both paths.

## Verification

- `pnpm --filter @paperclipai/server test --
heartbeat-plugin-environment issues-service`
- Manual QA: assign an issue to an agent that has a non-default
`defaultEnvironmentId`, confirm the issue's workspace settings now
include that
environment id without operator intervention. Trigger a rerun on an
issue
whose existing workspace points at a stale environment, confirm the
rerun uses
  the freshly-resolved environment.

## Risks

- Behavioural shift on assignment: previously assigning an agent didn't
propagate the agent's default environment to the issue. Now it does.
Callers
that explicitly want the issue to keep its existing/null environment
must set
`executionWorkspaceSettings.environmentId` themselves; the new logic
only
  fires when no explicit value is set.
- Behavioural shift on rerun: stale workspace config is no longer
applied to
  fresh runs. Operators who relied on this implicit inheritance may see
different environment selection on the first rerun after deploy.
Mitigation:
the explicit isssue settings and project policy are still honored as
before.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A (no UI changes)
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 18:33:55 -07:00
Devin Foley
09eceb952a Avoid resuming stale remote sessions (Pi adapter) (#5120)
> **Stacked PR (part 7 of 7).** Depends on:
  - PR #5114
  - PR #5115
  - PR #5116
  - PR #5117
  - PR #5118
  - PR #5119
> Diff against `master` includes commits from earlier PRs in the stack —
the new commit in this PR is the topmost one.

## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - The Pi adapter persists a session jsonl per agent so subsequent runs
resume
>   conversation context instead of starting cold
> - SSH testing reproduced a real failure: a verification issue reached
terminal
>   `done` and the agent claimed success, but the proof artifact
> `manual-qa/environment-matrix/ssh/pi_local.md` was missing from the
realized
>   SSH workspace on the QA target box
> - Root cause: the saved session header recorded a different cwd than
the new
> execution cwd, but the resume eligibility check only compared
session-params
> cwd via local-style `path.resolve` (which doesn't roundtrip on remote
POSIX
> paths). The stale session got resumed and writes landed in the wrong
cwd
> - This PR tightens resume eligibility for remote targets: it adds
remote-aware
> cwd normalisation, reads the first line of the session jsonl over SSH
(`head
>   -n 1`) to verify the saved header cwd, and only resumes when both
> session-params cwd *and* the on-disk header cwd match the realised
execution
>   cwd. Stale sessions are skipped silently and the run starts cold
> - The benefit is that Pi runs across cwd-changing environments stop
> accidentally resuming each other's sessions, and proof artifacts land
where
>   reviewers expect them

## What Changed

- Added `normalizeExecutionCwd`, `executionCwdsMatch`,
`readSessionHeaderCwd`,
  and `readSavedSessionCwd` helpers in `pi-local/src/server/execute.ts`
- `readSavedSessionCwd` reads the first line of the session jsonl —
locally via
`fs.readFile`, remotely via `runAdapterExecutionTargetShellCommand`
(`head -n 1`)
- Resume eligibility now requires:
  1. Saved session id is non-empty
  2. Execution target shape matches (existing check)
  3. Session-params cwd matches the realised execution cwd
4. Session-header cwd (from the on-disk jsonl) matches the realised
execution cwd
- Stale sessions are skipped silently (run starts cold) instead of
resumed
- `execute.remote.test.ts` extended with: matching header → resume;
mismatched
header → start fresh; missing/unreadable header → start fresh; remote
head
  command failure → start fresh

## Verification

- `pnpm --filter @paperclipai/adapter-pi-local test`
- `pnpm test -- pi-local`
- Manual QA: ran a Pi agent twice in two different remote cwds,
confirmed
the second run did not pick up the first run's session and that
subsequent
  runs in the original cwd still resumed correctly

## Risks

- Adds a `head -n 1` shell call per Pi run on remote targets. Negligible
  latency (single read of session jsonl), bounded by 15s timeout.
- If the `head` call fails for unrelated reasons (transient remote
unreachability), the run will start cold instead of resuming. This is
the
safe default but worth noting — operators may see one extra cold run if
a
  remote glitches mid-session.
- No data is deleted or migrated; stale sessions remain on disk for
manual
  inspection if desired.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 13:51:38 -07:00
Devin Foley
d22e790bd4 Validate remote model probes on execution target (OpenCode) (#5119)
> **Stacked PR (part 6 of 7).** Depends on:
  - PR #5114
  - PR #5115
  - PR #5116
  - PR #5117
  - PR #5118
> Diff against `master` includes commits from earlier PRs in the stack —
the new commit in this PR is the topmost one.

## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - The OpenCode adapter validates that its configured model exists
before letting
>   a run start so misconfiguration fails fast with a clear error
> - SSH testing reproduced an OpenCode failure where issues stayed
`backlog`,
>   timed out, and produced no comments. The root cause was in
> `packages/adapters/opencode-local/src/server/execute.ts`: the local
model
> guard `ensureOpenCodeModelConfiguredAndAvailable(...)` only ran when
execution
> was *not* remote, so SSH OpenCode bypassed it and failed silently
later
> - Subsequent testing surfaced a related remote-only failure where the
probe
> (when wired up naively) hits `EACCES: permission denied, mkdir
'/var/folders'`
> on the SSH box because of how OpenCode's runtime config picks a
tempdir
> - This PR runs the model probe on the actual execution target —
`opencode
> models` via `runAdapterExecutionTargetProcess` — instead of the local
CLI,
> parses the output with the shared `parseOpenCodeModelsOutput` helper,
and
> reports a concrete error naming the offending model and a sample of
available
>   remote models when the configured model isn't present
> - The benefit is that mismatched OpenCode models surface as a clear
pre-flight
> error referencing the remote target instead of a silent run that never
leaves
>   `backlog`

## What Changed

- Added `ensureRemoteOpenCodeModelConfiguredAndAvailable` in
  `opencode-local/src/server/execute.ts` that runs `opencode models` via
`runAdapterExecutionTargetProcess` and validates the configured model is
in
  the parsed output
- `models.ts` now exports `parseOpenCodeModelsOutput` and
`requireOpenCodeModelId`
  so the remote path can reuse them
- `execute.ts` calls the remote variant when `executionTargetIsRemote`,
otherwise
  the existing local `ensureOpenCodeModelConfiguredAndAvailable`
- Errors include the offending model id and a sample of available remote
models
  so the operator knows exactly what's missing
- `execute.remote.test.ts` extended with cases for: probe timeout, probe
  non-zero exit, empty model list, and missing-model error

## Verification

- `pnpm --filter @paperclipai/adapter-opencode-local test`
- `pnpm test -- opencode-local`
- Manual QA: configured an OpenCode agent with a model that exists
locally but
not in the remote sandbox, and confirmed the new error fires before the
run
  starts and references the remote target

## Risks

- New behaviour: remote model validation adds a `~20s timeout` `opencode
models`
call on every remote run start. For most environments this is fast, but
a
network-slow sandbox could see startup latency rise. Timeout is bounded.
- If the remote CLI is missing or misconfigured, the new error replaces
the old
generic startup failure — clearer message, but the failure point shifts
earlier. Monitor for any QA flows that relied on the old failure shape.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 13:34:09 -07:00
Devin Foley
856c6cb192 Fix remote workspace environment shaping (#5118)
> **Stacked PR (part 5 of 7).** Depends on:
  - PR #5114
  - PR #5115
  - PR #5116
  - PR #5117
> Diff against `master` includes commits from earlier PRs in the stack —
the new commit in this PR is the topmost one.

## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agents run with a Paperclip-shaped environment
(`PAPERCLIP_WORKSPACE_CWD`,
> worktree path, `PAPERCLIP_WORKSPACES_JSON` hints) so the CLI can
locate the
>   correct project tree
> - SSH testing reproduced a real failure: a Codex SSH run wrote to
> `/tmp/paperclip-env-matrix-...` (the *host* path) instead of the
realized
> remote workspace at `/home/<user>/paperclip-env-matrix-ssh-claude/...`
> because the adapter injected `PAPERCLIP_WORKSPACE_CWD=/tmp/...` into
the
>   remote env
> - Code review on the initial codex-only fix asked to roll the same
approach
> into every other SSH-capable adapter (claude, acpx, cursor, opencode,
gemini,
>   pi) via a shared helper rather than duplicating per-adapter
> - This PR adds `shapePaperclipWorkspaceEnvForExecution` in
adapter-utils that,
> when the execution target is remote: replaces local cwd with the
realized
> execution cwd, nulls out worktree path (which has no remote meaning),
and
> rewrites/strips `cwd` entries in workspace hints based on what was
actually
>   synced. Every adapter calls it before invoking the remote runner
> - The benefit is that remote runs see the realized remote workspace,
host-local
> paths stop leaking into remote env, and the rule is unit-tested in one
place

## What Changed

- Added `shapePaperclipWorkspaceEnvForExecution` to
  `packages/adapter-utils/src/server-utils.ts` with full unit coverage
  (`server-utils.test.ts`)
- Each of acpx-local, claude-local, codex-local, cursor-local,
gemini-local,
opencode-local, pi-local now calls the new shaper before issuing the
remote
  command and feeds the shaped values into `applyPaperclipWorkspaceEnv`
- Per-adapter `execute.remote.test.ts` files extended to cover the new
shaping
  behaviour: localhost paths replaced with remote cwd, foreign-cwd hints
  stripped, worktree path nulled out for remote targets
- `acpx-local/src/server/execute.test.ts` extended with shaping coverage

## Verification

- `pnpm test -- server-utils execute.remote`
- `pnpm --filter @paperclipai/adapter-acpx-local test`
- Manual QA reproducing the original failure:
  1. Provision an E2B sandbox environment for the Paperclip QA company
2. Assign an issue to a remote-targeted claude-local agent and confirm
the
run starts in the correct remote cwd (no `/Users/...` path leakage in
the
     run logs)
  3. Repeat for opencode-local and pi-local

## Risks

- Behavioural shift: hints whose `cwd` doesn't match the workspace cwd
are now
stripped on remote targets. If any adapter relied on a leaked local hint
cwd,
it will see a missing `cwd` instead. Reviewed all current callers — none
do.
- Adds a small per-run cost (path resolve + string normalisation) on
every remote
  execution. Negligible.
- Worktree path is now nulled out on remote (it has no meaning there).
Adapters
  that previously read the value defensively will continue to work.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 13:17:52 -07:00
Devin Foley
bb7d040894 Switch OpenCode to explicit static/local-aware model selection (#5117)
> **Stacked PR (part 4 of 7).** Depends on:
  - PR #5114
  - PR #5115
  - PR #5116
> Diff against `master` includes commits from earlier PRs in the stack —
the new commit in this PR is the topmost one.

## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - When creating an OpenCode-local agent, Paperclip currently validates
> `adapterConfig.model` against the *Paperclip host's* `opencode models`
output
> - SSH testing surfaced that this blocks creating an OpenCode agent for
an SSH
> environment: the model that exists on the SSH target isn't visible to
the
> host, so creation fails with "OpenCode requires `adapterConfig.model`
in
> provider/model format" even when the operator picked a real remote
model
> - The initial direction was environment-aware model discovery; the
final
> decision was to keep OpenCode on the same explicit-model pattern as
other
> adapters (default + curated list + manual override) and stop blocking
>   creation on host-side discovery
> - This PR does both: the adapter-models endpoint now accepts
`environmentId` and
> probes against the target environment, and the create-time hard gate
is
> replaced by `requireOpenCodeModelId` which validates `provider/model`
*format*
> without requiring host-local discovery. Test/run-time still surfaces
real
>   auth/availability problems
> - The benefit is that operators can create OpenCode agents for remote
> environments without out-of-band setup, and the model picker in the UI
>   reflects the actually-targeted environment

## What Changed

- Added `requireOpenCodeModelId(input)` in
`opencode-local/src/server/models.ts`,
  exported it from the adapter index
- `ensureOpenCodeModelConfiguredAndAvailable` now delegates the format
check to
  `requireOpenCodeModelId`
- `agentsApi.adapterModels(companyId, adapterType, { environmentId })`
now accepts
  an environment ID and passes it as a query parameter
- `queryKeys.agents.adapterModels` now keys on `(companyId, adapterType,
environmentId)`
- `server/src/routes/agents.ts` reads and validates the new query
parameter,
  forwarding it to the adapter's model probe
- `AgentConfigForm.tsx` and `OnboardingWizard.tsx` build the model query
key from
the currently selected default environment ID and disable autodetect for
  `opencode_local` (model selection is explicit)
- `NewAgent.tsx` simplified — no longer special-cases OpenCode
autodetect
- `company-portability.ts` no longer needs OpenCode-specific autodetect
handling
- Tests added/updated:
  `adapter-model-refresh-routes.test.ts`, `adapter-models.test.ts`,
`agent-permissions-routes.test.ts`,
`opencode-local/src/server/models.test.ts`

## Verification

- `pnpm --filter @paperclipai/server test -- adapter-models
adapter-model-refresh agent-permissions`
- `pnpm --filter @paperclipai/adapter-opencode-local test`
- `pnpm --filter @paperclipai/ui test -- AgentConfigForm
OnboardingWizard NewAgent`
- Manual QA in browser:
1. Boot Paperclip on Tailscale-bound port (so it's reachable from
another
machine), create an OpenCode-local agent, switch the default environment
between two installed sandboxes, and confirm the model list refreshes
     per-environment
  2. Submit with a malformed `provider/model` string and verify the new
     `requireOpenCodeModelId` error surfaces
- Before/after screenshots attached for `AgentConfigForm` model picker

## Risks

- Behavioural shift: switching default environment now triggers a model
refetch.
Should be cheap but introduces a new UI loading state for OpenCode
users.
- Removing dynamic autodetect for OpenCode: if any user configured an
agent
without specifying `model` and relied on autodetect populating it, that
agent
will now fail at submit time. Mitigation: validation error is explicit
and
  actionable.
- New query string parameter on `/api/companies/:id/adapter-models` —
older
clients that omit it still work (parameter is optional and defaults to
null).

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 13:01:34 -07:00
Devin Foley
076067865f Migrate SSH environment callback to bridge (#5116)
> **Stacked PR (part 3 of 7).** Depends on:
  - PR #5114
  - PR #5115
> Diff against `master` includes commits from earlier PRs in the stack —
the new commit in this PR is the topmost one.

## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agents executing on a remote SSH-backed environment need a way to
call back into
>   the Paperclip control plane (run events, log streaming, signals)
> - When the SSH host can't reach the Paperclip host (NAT, firewalls, or
simply not
> on the same network), the run silently fails or hangs — a recurring
class of
>   failure during SSH testing
> - In sandboxed environments we already solved this with a callback
bridge that
> tunnels back through the existing connection; SSH was the odd one out
> - This PR migrates SSH execution to use the same callback bridge, so
every
> adapter's remote run uses one consistent reverse-channel. Per-adapter
SSH glue
> is deleted in favour of a shared `CommandManagedRuntimeRunner` built
from the
>   SSH spec
> - The benefit is fewer SSH-specific failure modes, a smaller code
surface, and
>   one place to evolve the callback contract going forward

## What Changed

- Added `createSshCommandManagedRuntimeRunner` in
`packages/adapter-utils/src/ssh.ts` that adapts an SSH spec into a
generic
  command-managed-runtime runner (with cwd, env, and timeout handling)
- Removed `paperclipApiUrl` from `SshRemoteExecutionSpec`; the bridge
URL now flows
  through the shared runner
- Reworked `execution-target.ts` to use the SSH runner alongside sandbox
runners
  via a unified `CommandManagedRuntimeRunner` interface
- Simplified `remote-managed-runtime.ts` and
`sandbox-managed-runtime.ts` to consume
  the shared runner abstraction
- Deleted per-adapter SSH callback wiring from claude-local,
codex-local,
  cursor-local, gemini-local, opencode-local, pi-local execute.ts files
- Removed `environment-runtime-driver-contract.test.ts` (the contract is
now
  enforced by `environment-execution-target.test.ts`)
- Added/updated `execute.remote.test.ts` cases for each adapter to cover
the SSH
  runner path

## Verification

- `pnpm --filter @paperclipai/adapter-utils test`
- `pnpm test -- execute.remote` (covers all six local adapters' SSH
paths)
- Manual QA: ran a claude-local agent against an SSH-backed environment,
confirmed
the agent successfully called back to `/api/agent-callback/*` endpoints
during
  the run

## Risks

- Refactor touches all six local adapters. If any adapter had subtle
SSH-specific
behaviour that wasn't captured in tests, it could regress. Mitigation:
each
  adapter's `execute.remote.test.ts` was extended.
- `paperclipApiUrl` removal from `SshRemoteExecutionSpec` is a breaking
type change
for any internal consumer. Verified no external plugins consume this
type.
- The new `CommandManagedRuntimeRunner` shape is a public surface in
`@paperclipai/adapter-utils`; downstream plugins implementing custom
runners may
  need updates, but no such plugins exist in this repo.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 12:43:52 -07:00
Devin Foley
a7b45938b7 Let sandbox providers declare shell defaults (#5114)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agents execute in sandboxed remote environments served by pluggable
sandbox
>   providers (E2B today, more later)
> - Today every sandbox command runs under `sh -lc` regardless of what
the
>   provider's container actually ships
> - That misses bash-only shell init on E2B (which ships bash) and
prevents
> future providers from declaring a different default — there's no way
for a
>   provider to say "I have bash, use it"
> - This PR adds a `shellCommand` field to sandbox execution targets so
providers
> can declare their preferred shell ("bash" for E2B), threads it through
the
> sandbox-managed-runtime client, callback bridge, and execution-target
shell
>   helper, and validates the value at the lease-metadata boundary
> - The benefit is that sandbox commands run under the right shell on
the right
> provider, and adding new sandbox providers only needs to declare a
shell
>   preference

## What Changed

- Added `packages/adapter-utils/src/sandbox-shell.ts` exporting
`preferredShellForSandbox(shellCommand)` (returns `"bash"` if input is
`"bash"`,
  else `"sh"`)
- Added `shellCommand?: "bash" | "sh" | null` to
`AdapterSandboxExecutionTarget`
  and `CommandManagedRuntimeSpec`; threaded it through
`runAdapterExecutionTargetShellCommand`,
`prepareAdapterExecutionTargetRuntime`,
  and `startAdapterExecutionTargetPaperclipBridge`
- `createCommandManagedRuntimeClient`, `prepareCommandManagedRuntime`,
and
`createCommandManagedSandboxCallbackBridgeQueueClient` now take an
optional
  `shellCommand` and use `preferredShellForSandbox` to pick the shell
- `startSandboxCallbackBridgeServer` accepts a `shellCommand` for its
server
  startup, readiness probe, and stop hook
- E2B sandbox plugin declares `shellCommand: "bash"` in `leaseMetadata`
- `resolveEnvironmentExecutionTarget` reads `shellCommand` from lease
metadata
  (validating against `"bash" | "sh" | null`)
- `environment-runtime.ts` adds `"shellCommand"` to
`INTERNAL_PLUGIN_SANDBOX_CONFIG_KEYS`
so the field round-trips through internal plugin config without leaking
to
  external plugin metadata
- Updated tests in `command-managed-runtime.test.ts`,
  `execution-target-sandbox.test.ts`, `sandbox-callback-bridge.test.ts`,
  `environment-execution-target.test.ts`

## Verification

- `pnpm --filter @paperclipai/adapter-utils test`
- `pnpm --filter @paperclipai/server test --
environment-execution-target`
- `pnpm --filter @paperclipai/sandbox-providers-e2b test`
- Manual QA: boot a Paperclip instance, create an E2B-backed
environment, run a
claude_local agent against it, and confirm the run completes (verifies
bash
  shell semantics flow through the callback bridge end-to-end)

## Risks

- E2B sandbox commands now run under `bash -lc` instead of `sh -lc`.
Bash is a
strict superset for the commands we issue (no busybox-only flags in our
shell
scripts), so risk is low. The shellCommand field is opt-in via lease
metadata —
  providers that don't declare it stay on `sh`.
- New optional field on `CommandManagedRuntimeSpec` and
`AdapterSandboxExecutionTarget`.
  Consumers ignoring the field retain previous behaviour (sh).
- Lease metadata now carries an additional field. Existing leases
without
`shellCommand` resolve to `null` and fall back to sh — backwards
compatible.

## Model Used

- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A (no UI changes)
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-05-03 12:19:35 -07:00
Dotta
15eac43b43 [codex] Retry max-turn exhausted heartbeats (#5096)
## Thinking Path

> - Paperclip orchestrates AI agents for autonomous companies, and
heartbeat execution is the control-plane loop that keeps assigned work
moving.
> - Max-turn exhaustion is a recoverable local-adapter stop condition
for Claude and Gemini agents when a run needs another heartbeat to
continue safely.
> - The previous behavior could leave max-turn continuation details hard
to inspect, and duplicate/stale continuation wakes could keep running
after issue state changed.
> - The adapter layer also needed to avoid trusting arbitrary
stdout/stderr text as scheduler control metadata.
> - This pull request adds bounded max-turn continuation scheduling,
visible retry state, structured stop metadata handling, and
stale/duplicate continuation guards.
> - The benefit is safer automatic continuation after max-turn stops,
clearer operator visibility, and fewer duplicate or stale agent runs.

## What Changed

- Replaces closed PR #4952, whose head repository was deleted.
- Rebases the recovered max-turn continuation branch onto current
`paperclipai/paperclip:master`.
- Adds max-turn continuation scheduling and retry-state plumbing for
heartbeat runs.
- Adds stale/duplicate continuation suppression when issue status,
ownership, or execution locks change.
- Normalizes Claude/Gemini max-turn detection around structured stop
metadata instead of unstructured stdout/stderr text.
- Surfaces max-turn continuation settings and retry visibility in the
board UI.
- Adds focused server, adapter, and UI tests for max-turn stop metadata,
retry scheduling, stale queued-run invalidation, adapter
parsing/execution, run ledger display, and agent config patching.

## Verification

- `pnpm install --no-frozen-lockfile` to refresh local dependencies
after rebasing onto current `master`.
- `pnpm run preflight:workspace-links && pnpm exec vitest run
server/src/__tests__/claude-local-adapter.test.ts
server/src/__tests__/claude-local-execute.test.ts
server/src/__tests__/gemini-local-adapter.test.ts
server/src/__tests__/gemini-local-execute.test.ts
server/src/__tests__/heartbeat-retry-scheduling.test.ts
server/src/__tests__/heartbeat-stale-queue-invalidation.test.ts
server/src/services/heartbeat-stop-metadata.test.ts
ui/src/components/IssueRunLedger.test.tsx
ui/src/lib/agent-config-patch.test.ts ui/src/lib/runRetryState.test.ts
--testTimeout=20000`
- `pnpm --filter @paperclipai/adapter-claude-local typecheck && pnpm
--filter @paperclipai/adapter-gemini-local typecheck && pnpm --filter
@paperclipai/server typecheck && pnpm --filter @paperclipai/ui
typecheck`
- UI screenshot note: the UI changes are limited to config/ledger state
rendering rather than layout changes; component/unit coverage above
verifies the rendered behavior.

## Risks

- Medium behavior risk: heartbeat retry gating now suppresses max-turn
continuations when issue state or execution locks drift, so any callers
that relied on stale continuations running will now see cancellation
instead.
- Low adapter risk: Claude/Gemini unstructured text no longer triggers
max-turn scheduler metadata, so only structured stop signals and Gemini
exit code 53 are trusted.
- No database migrations.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex coding agent, GPT-5-class model, tool-enabled local
repository editing and command execution.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots (not applicable: state/default rendering only; covered by
component/unit tests)
- [x] I have updated relevant documentation to reflect my changes (not
applicable: no user-facing command or docs contract changed)
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-03 11:30:48 -05:00
Dotta
57229d0f24 [codex] Add issue monitor liveness controls (#4988)
## Thinking Path

> - Paperclip is a control plane for autonomous AI companies where work
must stay observable, governable, and recoverable.
> - The task/heartbeat subsystem owns agent execution continuity, issue
state transitions, and visible recovery behavior.
> - Waiting on an external service is not the same as being blocked when
the assignee still owns a future check.
> - The gap was that agents had no first-class one-shot monitor state
for external-service waits, so recovery could look stalled or require ad
hoc comments.
> - This pull request adds bounded issue monitors that can wake the
owner, clear exhausted waits, and produce explicit recovery behavior.
> - It also surfaces monitor status in the board UI and documents when
to use monitors versus `blocked`.
> - The benefit is clearer liveness semantics for asynchronous waits
without weakening single-assignee task ownership.

## What Changed

- Added issue monitor fields, shared types, validators, constants, and
an idempotent `0075` migration for scheduled monitor state.
- Added server-side monitor scheduling, dispatch, recovery bounds,
activity logging, and external-ref redaction.
- Added board/agent route coverage for monitor permissions and child
monitor scheduling.
- Added issue detail/property UI for monitor state, a monitor activity
card, and Storybook stories for review surfaces.
- Documented monitor semantics and recovery policy behavior in
`doc/execution-semantics.md`.
- Addressed Greptile review feedback by preserving monitor state in
skipped-stage builders and making board monitor saves send `scheduledBy:
"board"`.

## Verification

- `pnpm install --frozen-lockfile`
- `pnpm run preflight:workspace-links && pnpm exec vitest run
server/src/__tests__/issue-execution-policy-routes.test.ts
server/src/__tests__/issue-execution-policy.test.ts
server/src/__tests__/issue-monitor-scheduler.test.ts
server/src/__tests__/recovery-classifiers.test.ts
ui/src/components/IssueMonitorActivityCard.test.tsx
ui/src/components/IssueProperties.test.tsx
ui/src/lib/activity-format.test.ts`
- First run passed 5 files and failed to collect 2 server suites because
the worktree was missing the optional `acpx/runtime` dependency.
- After `pnpm install --frozen-lockfile`, reran the 2 failed suites
successfully.
- `pnpm exec vitest run
server/src/__tests__/issue-monitor-scheduler.test.ts
server/src/__tests__/recovery-classifiers.test.ts`
- `pnpm --filter @paperclipai/shared typecheck && pnpm --filter
@paperclipai/db typecheck && pnpm --filter @paperclipai/server typecheck
&& pnpm --filter @paperclipai/ui typecheck`
- `pnpm exec vitest run
server/src/__tests__/issue-execution-policy.test.ts
ui/src/components/IssueProperties.test.tsx`
- `pnpm --filter @paperclipai/server typecheck && pnpm --filter
@paperclipai/ui typecheck`
- `pnpm exec vitest run
ui/src/components/IssueMonitorActivityCard.test.tsx
ui/src/components/IssueProperties.test.tsx`
- `pnpm --filter @paperclipai/ui typecheck`
- Storybook screenshot captured from
`http://127.0.0.1:6006/iframe.html?viewMode=story&id=product-issue-monitor-surfaces--monitor-surfaces`
with Playwright.

## Screenshots

![Issue monitor Storybook
surfaces](https://raw.githubusercontent.com/paperclipai/paperclip/PAP-2945-when-a-task-is-waiting-for-an-_external-service_-what-state-should-it-be-in-and-what-recovery-method-could-it-h/docs/pr-screenshots/pap-2945/monitor-surfaces.png)

## Risks

- Medium: this changes heartbeat recovery behavior for scheduled
external-service waits, so regressions could affect wake timing or
recovery issue creation.
- Migration risk is reduced by using `IF NOT EXISTS` for the new issue
monitor columns and index.
- External monitor references are treated as secret-adjacent and are
intentionally omitted from visible activity/wake payloads.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent with repository tool use and terminal
execution.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots or Storybook review surfaces
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-03 08:58:53 -05:00
Dotta
76f09c8eb6 [PAP-3180] Move workspace switcher into sidebar (#4981)
## Thinking Path

> - Paperclip is the control plane for autonomous AI companies.
> - The board UI needs a clear persistent way to move between company
workspaces.
> - The previous layout kept company switching in a separate left rail,
which made the sidebar feel split between workspace selection and
navigation.
> - The workspace switcher belongs in the sidebar header so navigation
and workspace context stay together.
> - This pull request removes the separate company rail from the layout
and turns the sidebar company menu into the primary workspace switcher.
> - The benefit is a cleaner sidebar structure that keeps workspace
identity, switching, company actions, and navigation in one place.

## What Changed

- Removed the standalone `CompanyRail` from the main layout.
- Added the company/workspace switcher to the default, company settings,
and instance settings sidebars.
- Expanded `SidebarCompanyMenu` to list active workspaces, indicate the
current workspace, navigate out of instance settings when switching, and
expose add-company onboarding.
- Updated focused component tests for the new workspace-switcher
behavior.

## Verification

- `pnpm --filter @paperclipai/ui exec vitest run
src/components/SidebarCompanyMenu.test.tsx
src/components/CompanySettingsSidebar.test.tsx`
- `pnpm --filter @paperclipai/ui typecheck`
- `git diff --check`
- Visual smoke attempted against the managed dev server at
`http://127.0.0.1:57385`; a fresh browser context reached the
authenticated sign-in screen, so I could not capture an authenticated
sidebar screenshot from this heartbeat.

## Risks

- Low-to-medium UI risk: this changes the primary sidebar structure and
workspace-switching entry point.
- The instance-settings switch behavior now routes back to the selected
company dashboard when a workspace is selected.
- No migrations, API contracts, or lockfile changes.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool-enabled, medium reasoning mode.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-02 08:13:53 -05:00
Dotta
685ee84e4a [codex] Document terminal bench dispatch config (#4961)
## Thinking Path

> - Paperclip agents rely on skills for repeatable operating procedures
> - The Terminal-Bench loop skill needs to preserve enough dispatch
configuration to reproduce real heartbeat behavior
> - A bare benchmark command can create unassigned work with no
heartbeat-enabled agent, which is a harness setup failure rather than
product evidence
> - The Paperclip heartbeat skill also needs to keep escalation biased
toward agent-owned follow-through
> - This pull request documents dispatch runner config requirements and
strengthens the agent follow-through rule
> - The benefit is fewer misleading benchmark loops and clearer agent
operating guidance

## What Changed

- Documented `PAPERCLIP_HARBOR_RUNNER_CONFIG` / runner dispatch config
as required Terminal-Bench loop input.
- Updated the Terminal-Bench loop smoke check to require the dispatch
config mention.
- Added stronger Paperclip skill guidance to avoid asking humans for
work an agent can perform.

## Verification

- `pnpm smoke:terminal-bench-loop-skill`

## Risks

- Low risk: documentation and smoke expectation changes only. The
stricter smoke assertion is intentional so future edits do not drop the
dispatch config requirement.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 12:00:47 -05:00
Dotta
d7719423e9 [codex] Harden non-system database backup schemas (#4960)
## Thinking Path

> - Paperclip is a control plane whose database is the durable audit and
work record
> - Database backup needs to include operator/plugin schemas while
excluding PostgreSQL-owned internals
> - PostgreSQL reserves the `pg_` schema prefix for system schemas,
including temp and toast variants
> - A single escaped `pg_` prefix predicate is less brittle than
enumerating individual `pg_toast` and `pg_temp` forms
> - This pull request tightens non-system schema discovery for logical
backups without changing the normal user/plugin schema path

## What Changed

- Replaced narrow `pg_toast` and `pg_temp` schema exclusions with an
escaped `pg_` reserved-prefix exclusion.
- Kept `information_schema` excluded from logical backup metadata
discovery.
- Addressed Greptile feedback by removing redundant no-op additions from
the prior iteration.

## Verification

- `pnpm exec vitest run packages/db/src/backup-lib.test.ts`
- PR checks on the latest pushed head: policy, verify, e2e, Greptile
Review, and Snyk

## Risks

- Low risk: PostgreSQL reserves `pg_` schema names for system use, so
this should only exclude database-owned internals that should not be
restored from Paperclip logical backups.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected - check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 11:59:53 -05:00
Dotta
fe401b7fa9 [codex] Polish inbox nested issue UI (#4959)
## Thinking Path

> - Paperclip orchestrates AI agents through issue lists and
issue-thread interactions
> - The inbox must preserve nested issue visibility and keyboard
navigation as work decomposes into deeper sub-issues
> - Some UI polish issues made nested rows harder to scan and pending
question cancellation less covered
> - The issue list also had a small test indentation regression around
load-more behavior
> - This pull request tightens nested inbox rendering and related
issue-thread/list polish
> - The benefit is a more reliable operator inbox for multi-level work
trees

## What Changed

- Included nested grandchild issues in inbox keyboard navigation and
recursive row rendering.
- Sort parent rows by descendant activity so active subtrees remain
visible.
- Removed extra inbox card background styling in favor of the page
surface.
- Added regression coverage for pending question cancellation.
- Cleaned up the issue-list load-more test indentation.

## Verification

- `pnpm exec vitest run ui/src/lib/inbox.test.ts
ui/src/components/IssueChatThread.test.tsx
ui/src/components/IssuesList.test.tsx`
- Screenshots were not captured in this PR split; the visible flow is
covered by focused component/helper tests and should get browser QA in
the follow-up issue.

## Risks

- Medium risk: nested inbox rendering and keyboard navigation are
user-visible. The changes are localized to inbox grouping/rendering
helpers and covered by targeted tests.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 11:58:53 -05:00
Dotta
2d72292ad6 [codex] Add workspace routine run tab (#4958)
## Thinking Path

> - Paperclip orchestrates AI agents through reusable execution
workspaces and routines
> - Operators need a fast way to run workspace-aware routines against a
specific execution workspace
> - The existing workspace detail surface showed configuration, runtime
logs, and linked issues, but not routines that depend on workspace
variables
> - Routine runs also needed to prefill the selected execution workspace
so branch variables resolve correctly
> - This pull request adds a workspace routines tab and prefilled
routine-run dialog support
> - The benefit is a tighter workflow for rerunning reviews, smoke
checks, and other workspace-specific routines

## What Changed

- Added an execution workspace `Routines` tab and company-prefixed
routes.
- Listed routines that declare or reference workspace-specific
variables.
- Added `Run now` support that preselects the current execution
workspace in `RoutineRunVariablesDialog`.
- Centralized reusable execution workspace ordering/deduplication for
issue creation and workspace cards.
- Added focused UI helper and dialog regression tests.

## Verification

- `pnpm exec vitest run ui/src/lib/reusable-execution-workspaces.test.ts
ui/src/lib/workspace-routines.test.ts
ui/src/components/RoutineRunVariablesDialog.test.tsx
ui/src/lib/company-routes.test.ts`
- Screenshots were not captured in this PR split; the visible flow is
covered by focused component/helper tests and should get browser QA in
the follow-up issue.

## Risks

- Medium risk: this adds a new workspace detail tab and routine-run
path. It is isolated to workspace-scoped routines and uses existing
routine run APIs.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 11:58:15 -05:00
Dotta
570a4206da [codex] Recover productive terminal continuations (#4956)
## Thinking Path

> - Paperclip orchestrates AI agents through issue-scoped heartbeat runs
> - Recovery logic decides whether in-progress work still has a live
path after a terminal run
> - A productive terminal continuation can still leave an issue stranded
when no active run or wake remains
> - Treating that state as healthy leaves work stuck despite evidence
that more action is needed
> - This pull request re-enqueues recovery for productive terminal
continuations that left no live path
> - The benefit is fewer silently stranded in-progress issues after
agents make partial progress

## What Changed

- Reclassified successful-but-productive terminal continuations as
recoverable when no live path remains.
- Enqueue a follow-up recovery wake with the original run id and
continuation metadata.
- Added regression tests covering productive terminal continuation
recovery and advanced liveness handoff.

## Verification

- `pnpm exec vitest run
server/src/__tests__/heartbeat-process-recovery.test.ts
server/src/__tests__/run-continuations.test.ts`

## Risks

- Medium risk: recovery may schedule one more follow-up where Paperclip
previously considered the work observed. The existing uniqueness,
budget, and escalation checks still constrain retry loops.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 11:57:23 -05:00
Dotta
3cd26a78fc [codex] Surface live run comment context (#4957)
## Thinking Path

> - Paperclip orchestrates AI agents through issue comments and
heartbeat runs
> - The board UI needs to distinguish a comment that triggered a live
run from comments queued after that run started
> - The run payload already stores comment context, but active-run API
responses did not expose the ids the UI needs
> - Without those ids, the triggering comment can flash as queued while
the agent is already responding to it
> - This pull request exposes live-run comment context and teaches the
optimistic comment helper to ignore the trigger comment
> - The benefit is clearer issue-chat state during comment-triggered
agent interruptions

## What Changed

- Added `contextCommentId` and `contextWakeCommentId` to active/live run
payloads.
- Threaded those ids through server routes, heartbeat summaries, UI API
types, and issue detail rendering.
- Updated optimistic comment classification to avoid marking the
triggering comment as queued.
- Added server and UI regression coverage.

## Verification

- `pnpm exec vitest run
server/src/__tests__/agent-live-run-routes.test.ts
ui/src/lib/optimistic-issue-comments.test.ts`

## Risks

- Low-to-medium risk: adds optional fields to existing run payloads.
Existing consumers should ignore unknown fields, and UI handling is
null-safe.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 10:44:11 -05:00
Dotta
e8275318ba [codex] Raise agent heartbeat concurrency default (#4954)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Agent heartbeat settings control how much parallel work one employee
can run
> - The previous default of 5 concurrent runs was too restrictive for
active local agent teams
> - The shared default, heartbeat clamp, docs, and route/import/UI
expectations need to agree
> - This pull request raises the default heartbeat concurrency to 20
while keeping explicit headroom up to 50 for power users
> - The benefit is higher throughput for agent teams without each new
agent needing manual runtime config edits

## What Changed

- Raised `AGENT_DEFAULT_MAX_CONCURRENT_RUNS` from 5 to 20.
- Raised the heartbeat service max clamp from 10 to 50, keeping the new
default below the ceiling.
- Updated V1 implementation docs and tests that assert default
imported/exported runtime config.
- Updated the new-agent UI runtime config test to assert the shared
default constant instead of duplicating the numeric value.

## Verification

- `pnpm exec vitest run
server/src/__tests__/agent-permissions-routes.test.ts
server/src/__tests__/company-portability.test.ts
ui/src/lib/new-agent-runtime-config.test.ts`

## Risks

- Medium risk: new agents can consume more local execution capacity by
default. The heartbeat scheduler still respects configured max
concurrency and budget/pause controls, and operators can lower or raise
the per-agent cap within the `1..50` clamp.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5 coding agent, tool use and local command
execution. Exact context window was not exposed in the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 10:42:56 -05:00
Dotta
e273d621fc [PAP-3154] Stop padding /live-runs by default (#4963)
## Summary
- Fix [PAP-3154](/PAP/issues/PAP-3154): the Sidebar's "Dashboard NN
live" badge showed a constant 50 in every company because `GET
/api/companies/:companyId/live-runs` was padding its response with up to
50 recent (non-live) heartbeat runs whenever the caller did not pass
`minCount`.
- Regression introduced by
[#4875](https://github.com/paperclipai/paperclip/pull/4875) (commit
`6445bef9`), which capped both `minCount` and `limit` at 50 with a
fallback of 50 for omitted values. The cap is correct for `limit` (real
unboundedness guard); for `minCount` it conflates "no padding" with "pad
to the cap".
- Default `minCount` to 0 so callers asking for "live runs" only get
actually-live runs unless they explicitly request padding
(`ActiveAgentsPanel` is the only caller that does). Keep `limit` capped
at 50 by default.

## Test plan
- [x] `pnpm exec vitest run
server/src/__tests__/agent-live-run-routes.test.ts` — 7/7 pass,
including new tests for the no-pad default and explicit padding.
- [x] `pnpm exec vitest run ui/src/components/Sidebar.test.tsx
ui/src/components/ActiveAgentsPanel.test.tsx
ui/src/api/heartbeats.test.ts` — 6/6 pass.
- [ ] Verify in dev: with ~8 truly-live runs in a company, the sidebar
Dashboard badge shows the real count (not 50).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 10:33:13 -05:00
Dotta
42a299fb9d [codex] Bound productivity review recovery loops (#4948)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies.
> - The heartbeat/productivity review subsystem detects when assigned
work is likely stuck or churning.
> - Productivity reviews are useful, but repeated reconciliation can
create noisy refresh comments or repeated review issues around the same
source issue.
> - That makes manager follow-up harder because the signal can get
buried under duplicate review activity.
> - This pull request bounds productivity review refreshes and creation
loops while preserving the existing escalation path.
> - The benefit is a quieter recovery loop that still surfaces stuck or
high-churn work for manager attention.

## What Changed

- Added refresh throttling for open productivity review issues,
including a one-hour default interval and a maximum of three refresh
comments per open review.
- Added a rolling 24-hour creation cap so completed/closed reviews
cannot immediately recreate review issues indefinitely for the same
source issue.
- Excluded cancelled productivity reviews from the creation cap so
manager cancellations do not silently suppress future legitimate
reviews.
- Preserved productivity review timestamps in deterministic test paths
and added targeted coverage for immediate refresh suppression, refresh
caps, creation caps, and cancelled-review exclusion.

## Verification

- `pnpm run preflight:workspace-links && pnpm exec vitest run
server/src/__tests__/productivity-review-service.test.ts`
- `pnpm exec vitest run
server/src/__tests__/productivity-review-service.test.ts`
- Greptile Review: 5/5 on commit
`bcf25832d0ffae25890b2ee7eed112d1c2d114fe` with review threads resolved.
- GitHub PR checks passed on the latest head: `policy`, `verify`, `e2e`,
`Greptile Review`, and `security/snyk (cryppadotta)`.
- Verified the branch is rebased onto `public-gh/master` with no
conflicts.
- Verified the diff does not include `pnpm-lock.yaml`, database schema
changes, or migrations.

## Risks

- Low-to-medium risk: this changes automation cadence for productivity
reviews. A truly stuck issue may receive fewer repeated refresh
comments, but the original review issue remains open and assigned for
manager action.
- No migration risk: this is server logic and tests only.

> Checked [`ROADMAP.md`](ROADMAP.md) for overlapping planned core work;
this is a targeted recovery-loop fix and does not add a new roadmap
feature.

## Model Used

- OpenAI Codex coding agent, GPT-5 model family, tool-using software
engineering mode. Exact context window is not exposed in this runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots (not applicable; server-only change)
- [x] I have updated relevant documentation to reflect my changes (not
applicable; no user-facing docs or commands changed)
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-05-01 08:32:04 -05:00
Devin Foley
d2dd759caa plugins: make e2b template default explicit (#4901)
## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - Remote execution environments are part of that control plane,
including sandbox-provider plugins like E2B
> - The E2B provider already normalizes config and runtime behavior
around a `base` template default
> - But the manifest still presented `template` as required, which
forces redundant operator input and makes the UI contract stricter than
runtime behavior
> - That mismatch showed up while building a repeatable QA workflow for
sandbox testing
> - This pull request makes the manifest and validation contract line up
with the existing `base` default
> - The benefit is a simpler and more accurate E2B environment setup
experience

## What Changed

- Removed the E2B manifest's `required: ["template"]` requirement so the
config schema matches runtime behavior
- Clarified the manifest description to say the template defaults to
`base` when omitted
- Added a focused unit test proving that validation normalizes a missing
template to `base`

## Verification

- Ran the focused E2B plugin test for the new behavior:
- `cd packages/plugins/sandbox-providers/e2b && pnpm test --
--testNamePattern "defaults a missing template to base"`

## Risks

- Low risk. This only loosens the schema to match the plugin's existing
runtime normalization and adds a test for that path.
- The broader E2B plugin suite currently has unrelated existing failures
outside this change; this PR does not modify those paths.

## Model Used

- OpenAI Codex, GPT-5 Codex via Codex CLI agent tooling, large-context
coding workflow with terminal tool use and local test execution.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [ ] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [ ] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] I will address all Greptile and reviewer comments before
requesting merge
2026-04-30 22:43:24 -07:00
Devin Foley
b02e67cea5 fix(ci): diff PR workflow paths from merge base (#4903)
## Thinking Path

> - Paperclip’s PR workflow is part of the control-plane safety surface
because it decides whether a branch is allowed to merge.
> - This issue started in that workflow: the lockfile and manifest
policy checks were diffing `base.sha..head.sha`, which incorrectly
treated unrelated `master` commits as if they belonged to the PR branch.
> - The right fix there is to diff from the PR merge base
(`base...head`) so policy checks only evaluate files introduced by the
branch itself.
> - Once that workflow fix was in place, `/checkpr` exposed a second
blocker on the PR merge ref: `verify` was failing in newer `master`-side
tests that were not part of the original branch diff.
> - The actionable repeated failure came from the ACPX local adapter
test suite, where a test hard-coded the managed Codex home under
`instances/default` even though the stable Vitest runner sets a
non-default `PAPERCLIP_INSTANCE_ID`.
> - This pull request now includes both the original CI diff-scope fix
and the targeted ACPX test fix so the PR’s actual checks align with
current base-branch execution.
> - The benefit is that the original false-positive lockfile failure is
removed, and the merge-ref verify path is hardened against the
instance-id isolation used in CI.

## What Changed

- Updated `.github/workflows/pr.yml` so the lockfile policy and manifest
policy steps diff `pull_request.base.sha...pull_request.head.sha` from
the merge base instead of using a two-dot base/head diff.
- Added an inline workflow comment explaining why the three-dot diff is
required for PR-scoped file detection.
- Updated `packages/adapters/acpx-local/src/server/execute.test.ts` so
the managed Codex home assertion uses a test-specific
`PAPERCLIP_INSTANCE_ID` instead of hard-coding `default`.
- Restored `PAPERCLIP_INSTANCE_ID` after that ACPX test finishes so the
test remains isolated and does not leak process env changes.

## Verification

- Reproduced the original false positive locally by comparing PR heads
`#4901` and `#4902` with the old `base..head` logic; both incorrectly
included `pnpm-lock.yaml` from unrelated `master` commits.
- Verified the new `base...head` logic reduces those PRs to only their
actual changed files and excludes `pnpm-lock.yaml`.
- Verified a real manifest-changing PR (`#4893`) still reports
`package.json` changes under the new logic.
- Ran `pnpm -r typecheck` successfully.
- Ran `pnpm vitest run
packages/adapters/acpx-local/src/server/execute.test.ts` successfully
after the ACPX test fix.
- Ran `pnpm vitest run packages/db/src/backup-lib.test.ts` successfully
against the merge-ref-related DB failure path observed during
`/checkpr`.
- Pushed commit `9520a976` and allowed PR `#4903` checks to rerun on the
updated branch.

## Risks

- Low risk: the workflow change only affects how PR policy checks
determine the changed file set.
- Low risk: the ACPX change is test-only and aligns the test with the
instance-isolation behavior already used by
`scripts/run-vitest-stable.mjs` in CI.
- The remaining operational risk is limited to other unrelated
merge-ref-only failures that were not reproduced in the targeted local
verification above.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, `gpt-5-codex`, via the Codex local adapter in Paperclip.
- Tool-using coding model with shell execution, git, GitHub CLI, and
repository inspection in a local worktree.
- Context included the current repo, the Paperclip task thread, PR check
output, and the isolated execution workspace.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [ ] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-04-30 21:22:40 -07:00
github-actions[bot]
6a7cca95ef chore(lockfile): refresh pnpm-lock.yaml (#4899)
Auto-generated lockfile refresh after dependencies changed on master.
This PR only updates pnpm-lock.yaml.

Co-authored-by: lockfile-bot <lockfile-bot@users.noreply.github.com>
2026-04-30 20:00:07 -05:00
Dotta
4272c1604d Add ACPX local adapter runtime (#4893)
## Thinking Path

> - Paperclip orchestrates AI-agent companies through a control plane
that can start, supervise, and recover agent runs.
> - Local adapters are the bridge between Paperclip issues and concrete
agent runtimes such as Claude, Codex, and other ACP-compatible tools.
> - The roadmap calls out broader “bring your own agent” and claw-style
agent support, and ACPX gives Paperclip one path to normalize multiple
ACP agents behind a single adapter.
> - The branch needed to become one reviewable PR against current
`paperclipai/paperclip:master`, without carrying stale base conflicts or
generated lockfile churn.
> - This pull request adds an experimental built-in `acpx_local`
adapter, integrates it through the server/CLI/UI adapter surfaces, and
adds regression coverage for runtime execution, skill sync, stream
parsing, diagnostics, and log redaction.
> - The benefit is that Paperclip can run Claude/Codex/custom ACP agents
through ACPX while keeping operator configuration, skills, logging, and
transcript rendering inside the existing adapter model.

## What Changed

- Added `@paperclipai/adapter-acpx-local` with server execution, config
schema, ACPX session handling, CLI formatting, UI config helpers, and
stdout parsing.
- Registered `acpx_local` across CLI, server, shared constants, UI
adapter metadata, adapter capabilities, and agent creation/editing
surfaces.
- Added ACPX runtime execution support with persistent sessions,
local-agent JWT environment handling, skill snapshots, runtime skill
materialization, and isolation/security regressions.
- Added ACPX adapter diagnostics and marked the adapter experimental in
the UI.
- Added command/env secret redaction for resolved command metadata in
adapter-utils, server event storage, and the Agent Detail invocation UI.
- Added Storybook coverage for ACPX config, transcript rendering, and
skill states, plus PR screenshots under `docs/pr-screenshots/pap-2944/`.
- Rebased the branch onto current `public-gh/master`; `pnpm-lock.yaml`
is intentionally not included and there are no migration/schema changes.

## Verification

- `pnpm exec vitest run
packages/adapters/acpx-local/src/server/execute.test.ts
packages/adapters/acpx-local/src/server/test.test.ts
packages/adapters/acpx-local/src/cli/format-event.test.ts
packages/adapters/acpx-local/src/ui/parse-stdout.test.ts
packages/adapter-utils/src/server-utils.test.ts
server/src/__tests__/redaction.test.ts
server/src/__tests__/acpx-local-execute.test.ts
server/src/__tests__/acpx-local-skill-sync.test.ts
server/src/__tests__/acpx-local-adapter-environment.test.ts
server/src/__tests__/adapter-routes.test.ts
server/src/__tests__/agent-skills-routes.test.ts
ui/src/adapters/metadata.test.ts` — 12 files, 87 tests passed.
- `pnpm --filter @paperclipai/adapter-acpx-local typecheck` — passed.
- `pnpm --filter @paperclipai/server typecheck` — passed.
- `pnpm --filter @paperclipai/ui typecheck` — passed.
- Confirmed PR diff does not include `pnpm-lock.yaml`, database schema
files, or migrations.

Screenshots:

![ACPX Claude skills
light](https://github.com/cryppadotta/paperclip-1/blob/PAP-2944-acpx-make-a-claude_local-adapter-that-uses-acpx-instead/docs/pr-screenshots/pap-2944/skills-claude-light.png?raw=true)
![ACPX Claude skills
dark](https://github.com/cryppadotta/paperclip-1/blob/PAP-2944-acpx-make-a-claude_local-adapter-that-uses-acpx-instead/docs/pr-screenshots/pap-2944/skills-claude-dark.png?raw=true)
![ACPX custom skills
light](https://github.com/cryppadotta/paperclip-1/blob/PAP-2944-acpx-make-a-claude_local-adapter-that-uses-acpx-instead/docs/pr-screenshots/pap-2944/skills-custom-light.png?raw=true)

## Risks

- Medium risk: this introduces a new built-in adapter package and
touches runtime execution, adapter registration, agent config, skills,
and transcript rendering.
- ACPX and ACP agent behavior can vary by installed tool versions; the
adapter is marked experimental to set operator expectations.
- `pnpm-lock.yaml` is excluded per repository PR policy, so dependency
lock refresh must be handled by the repo’s automation or maintainers.
- No database migration risk: no schema or migration files changed.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex coding agent based on GPT-5, with repository tool use,
shell execution, git operations, and local verification. Exact hosted
context window was not exposed in this environment.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-04-30 19:57:05 -05:00
Dotta
ad5432fece [codex] Harden issue recovery reliability (#4875)
## Thinking Path

> - Paperclip is the control plane for autonomous agent companies, so
non-terminal issue state must always have a clear live, waiting, or
recovery owner.
> - This change stays inside the server reliability and liveness
subsystem for assigned issue recovery, blocker attention, and live-run
polling.
> - Closed PR #4860 mixed this reliability work with separate
mutation-boundary policy changes, which made review and merge risk too
broad.
> - [PAP-2981](/PAP/issues/PAP-2981) asked for a replacement PR
containing only the remaining reliability slice and explicitly excluding
user-assignment and execution-policy restrictions.
> - Follow-up review also split `advanced` run-liveness continuation
behavior out of this PR so it can be reviewed separately.
> - The implementation hardens repeated recovery escalation, expands
blocker-attention coverage for explicit waiting and recovery paths, and
caps company live-run polling defaults.
> - The benefit is a smaller reliability PR that improves liveness
behavior without changing agent/user mutation authorization boundaries
or `advanced` continuation semantics.

## What Changed

- Avoid repeated liveness escalation updates when the source issue is
already blocked by the same open escalation.
- Treat open liveness escalation recovery issues, their source issues,
and their leaf blockers as covered waiting paths in blocker attention.
- Cap default company live-run polling at 50 rows for both `minCount`
and `limit`, including explicit zero values, to avoid unbounded
responses.
- Preserve the existing behavior where succeeded `advanced` runs are
considered productive/healthy for stranded-work recovery and are not
actionable bounded run-liveness continuations.
- Added focused server coverage for recovery dedupe, blocker attention,
liveness escalation, run continuations, and live-run polling.

## Verification

- `pnpm install --frozen-lockfile`
- `pnpm exec vitest run
server/src/__tests__/heartbeat-process-recovery.test.ts
server/src/__tests__/heartbeat-issue-liveness-escalation.test.ts
server/src/__tests__/issue-blocker-attention.test.ts
server/src/__tests__/run-continuations.test.ts
server/src/__tests__/agent-live-run-routes.test.ts`
- Result: 5 files passed, 63 tests passed.
- `pnpm --filter @paperclipai/server typecheck`
- Result: passed.
- No UI changes; screenshots are not applicable.

## Risks

- Recovery and blocker-attention classification changes can affect which
blocked chains are shown as covered versus needing attention.
- Live-run polling now treats omitted, invalid, or non-positive `limit`
/ `minCount` values as the capped default of 50.
- `advanced` run-liveness continuation behavior is intentionally
excluded from this PR and split for separate review.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5, code execution and GitHub CLI tool use, medium
reasoning effort.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-04-30 16:44:28 -05:00
Dotta
a3de1d764d Add cheap model profiles for local adapters (#4881)
## Thinking Path

> - Paperclip is a control plane for autonomous AI companies, where
adapters are the boundary between the board, agents, and execution
runtimes.
> - Local adapters currently expose a primary runtime configuration, but
operators often need a cheaper model lane for routine or low-risk work.
> - That cheap lane has to stay adapter-owned: runtime profile settings
should not mutate the primary adapter config or bypass existing
auth/secret mediation.
> - Issue creation also needs an ergonomic way to request primary,
cheap, or custom model behavior for a selected assignee.
> - This pull request adds a first-class `cheap` model profile contract
across adapter capabilities, heartbeat config resolution, agent
configuration, and issue creation.
> - The benefit is cheaper task execution can be configured and
requested explicitly while preserving adapter boundaries, secret
handling, and audit visibility.

## What Changed

- Added adapter model-profile capability metadata and a `cheap` profile
contract for supported local adapters.
- Applied `runtimeConfig.modelProfiles.cheap.adapterConfig` during
heartbeat config resolution, including requested/applied/fallback run
metadata.
- Added agent configuration UI for cheap model profile settings without
writing those settings into primary `adapterConfig`.
- Added New Issue assignee model lane controls for Primary / Cheap /
Custom and request payload handling.
- Added run ledger profile badges and Storybook stories for the new
cheap-lane UI states.
- Added tests for validators, heartbeat model profile application,
permission/secret mediation, UI payload helpers, and run ledger
rendering.
- Added committed UI verification screenshots under
`docs/pr-screenshots/pap-2837/`.
- Addressed Greptile review feedback around cheap-profile defaults,
shared profile types, and fallback test data.

## Verification

Local:

- `pnpm exec vitest run packages/shared/src/validators/issue.test.ts
server/src/__tests__/adapter-registry.test.ts
server/src/__tests__/agent-permissions-routes.test.ts
server/src/__tests__/heartbeat-model-profile.test.ts
ui/src/components/IssueRunLedger.test.tsx
ui/src/lib/agent-config-patch.test.ts
ui/src/lib/issue-assignee-overrides.test.ts
ui/src/lib/new-agent-runtime-config.test.ts` — passed, 8 files / 103
tests.
- `pnpm exec vitest run ui/src/lib/new-agent-runtime-config.test.ts
ui/src/components/IssueRunLedger.test.tsx` — passed after
Greptile/rebase follow-up, 2 files / 17 tests.
- `pnpm --filter @paperclipai/ui typecheck` — passed after
Greptile/rebase follow-up.
- `pnpm -r typecheck` — passed.
- `pnpm build` — passed.
- `pnpm test:run` — did not complete successfully in this local
worktree: it stopped in pre-existing `@paperclipai/adapter-utils`
sandbox/SSH fixture suites outside this PR diff. Failures were 5s local
timeouts plus `git init -b` unsupported by this machine's Git 2.21.0.
The branch-specific targeted suites above passed.
- Branch was fetched/rebased onto `public-gh/master`; `git rev-list
--left-right --count public-gh/master...HEAD` reports `0 9`.

Remote PR checks on latest head
`e30bf399146451c86cee98ed528d51d33fa5af5a`:

- `policy` — passed.
- `verify` — passed.
- `e2e` — passed.
- `Greptile Review` — passed, confidence score 5/5; Greptile review
threads resolved.
- `security/snyk (cryppadotta)` — passed.

Screenshots:

- [New issue cheap lane
desktop](https://github.com/paperclipai/paperclip/blob/PAP-2837-plan-cheap-model-for-adapters-that-can-support-it/docs/pr-screenshots/pap-2837/newissue-cheap-desktop.png)
- [New issue custom lane
desktop](https://github.com/paperclipai/paperclip/blob/PAP-2837-plan-cheap-model-for-adapters-that-can-support-it/docs/pr-screenshots/pap-2837/newissue-custom-desktop.png)
- [New issue unsupported adapter
desktop](https://github.com/paperclipai/paperclip/blob/PAP-2837-plan-cheap-model-for-adapters-that-can-support-it/docs/pr-screenshots/pap-2837/newissue-unsupported-desktop.png)
- [Run ledger model profile badges
desktop](https://github.com/paperclipai/paperclip/blob/PAP-2837-plan-cheap-model-for-adapters-that-can-support-it/docs/pr-screenshots/pap-2837/runledger-profile-badges-desktop.png)
- Mobile variants are also in `docs/pr-screenshots/pap-2837/`.

## Risks

- Medium: heartbeat config mediation now merges runtime model profiles
into adapter configs, so adapter secret normalization and host-command
restrictions must keep covering nested config paths.
- Medium: the UI adds another issue creation choice; unsupported
adapters must keep hiding the cheap lane and preserve primary behavior.
- Low migration risk: no database migration is included.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

OpenAI Codex coding agent using GPT-5-class reasoning with repo tool use
and command execution. Exact served model/context window was not exposed
by the runtime.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [ ] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-30 15:32:04 -05:00
Dotta
1fe1067361 Polish board settings and skills workflow (#4863)
## Thinking Path

> - Paperclip's board UI and bundled skills are the operator layer for
configuring agents, routines, issue workflows, and local troubleshooting
loops.
> - The prior rollup mixed this operator polish with database backups,
backend reliability, thread scale, and cost/workflow primitives.
> - This pull request isolates the remaining board QoL, settings,
issue-detail integration, adapter config cleanup, and skills smoke
tooling.
> - It includes some integration-level overlap with the thread and
workflow slices so this branch can run from `origin/master` while still
preserving the full original work.
> - Preferred merge order is the narrower primitives first, then this
integration PR last.
> - The benefit is that reviewers can inspect the user-facing
board/settings/skills layer separately from backend infrastructure
changes.

## What Changed

- Added board/settings polish for agents, routines, company settings,
project workspace detail, and issue detail controls.
- Added agent/routine UI regression tests and New Issue dialog coverage.
- Integrated issue-detail activity/cost/interaction surfaces and leaf
work pause/resume controls.
- Cleaned bundled adapter UI config defaults and onboarding copy.
- Added terminal-bench loop and work-stoppage diagnosis skills plus a
smoke test script.
- Updated attachment type handling and Paperclip skill/API guidance.

## Verification

- `pnpm install --frozen-lockfile`
- `pnpm exec vitest run ui/src/pages/Agents.test.tsx
ui/src/pages/Routines.test.tsx ui/src/components/NewIssueDialog.test.tsx
ui/src/pages/IssueDetail.test.tsx
server/src/__tests__/costs-service.test.ts
server/src/__tests__/issue-thread-interaction-routes.test.ts
server/src/__tests__/issue-thread-interactions-service.test.ts`
- Result: 7 test files passed, 54 tests passed.
- `pnpm run smoke:terminal-bench-loop-skill`
- Result: JSON output included `"ok": true` and `"cleanup": true`.
- UI screenshots not included because verification is focused
component/page coverage for the changed board surfaces.

## Risks

- This is the integration-heavy PR in the split and intentionally
overlaps some component/API primitives with the issue-thread and
workflow PRs so it can run from `origin/master`.
- Preferred merge order: #4859, #4860, #4861, #4862, then this PR last.
If earlier branches merge first, this PR may need a straightforward
conflict refresh in shared UI files.
- The terminal-bench smoke script creates temporary mock issues and
relies on cleanup; the verified run returned `cleanup: true`.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5.5, code execution and GitHub CLI tool use, medium
reasoning effort.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-04-30 15:28:11 -05:00
Dotta
c4269bab59 Add workflow interaction cancellation and issue cost summaries (#4862)
## Thinking Path

> - Paperclip coordinates work through issue-thread interactions, run
history, and cost telemetry.
> - Operators need workflow prompts to be cancellable and costs to be
visible at the issue level.
> - The earlier rollup mixed this workflow/cost work with database
backups, reliability recovery, thread scaling, and settings polish.
> - This pull request isolates the interaction and cost surfaces into a
reviewable slice.
> - The backend now supports cancelling pending question interactions
and summarizing issue-tree costs.
> - The UI component layer can render cancelled questions and interleave
activity with run ledger rows.

## What Changed

- Added `cancelled` as an issue-thread interaction status and result
shape for question interactions.
- Added the board-only `POST
/issues/:id/interactions/:interactionId/cancel` route and service
implementation.
- Added issue-tree cost summary support in the cost service and
`/issues/:id/cost-summary` API route.
- Extended shared cost exports and UI API/query keys for issue cost
summaries.
- Updated `IssueThreadInteractionCard` and `IssueRunLedger` components
for cancelled questions, issue cost surfaces, and activity/run
interleaving.
- Added focused server and component regression coverage.

## Verification

- `pnpm install --frozen-lockfile`
- `pnpm exec vitest run server/src/__tests__/costs-service.test.ts
server/src/__tests__/issue-thread-interaction-routes.test.ts
server/src/__tests__/issue-thread-interactions-service.test.ts
ui/src/components/IssueRunLedger.test.tsx`
- Result: 4 test files passed, 45 tests passed.
- UI screenshots not included because this PR updates reusable
components and API surfaces without wiring a new page-level layout.

## Risks

- Adds a new interaction terminal status; clients that switch
exhaustively on interaction status may need to handle `cancelled`.
- Issue-tree cost summaries use recursive issue traversal and should be
watched on unusually large issue trees.
- Page-level issue detail wiring is intentionally left to the board
QoL/issue-detail branch to keep this PR narrow.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5.5, code execution and GitHub CLI tool use, medium
reasoning effort.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-04-30 13:57:25 -05:00
Dotta
87f19cd9a6 Improve issue thread scale and markdown polish (#4861)
## Thinking Path

> - Paperclip's board UI is the operator surface for supervising
AI-agent companies.
> - Issue threads are where operators read progress, respond to agents,
inspect markdown, and jump through long histories.
> - Large threads and rich markdown had become difficult to navigate and
expensive to render.
> - The previous rollup mixed these UI scale fixes with unrelated
backend recovery, costs, backups, and settings changes.
> - This pull request isolates the issue-thread scale and markdown
polish work.
> - The benefit is a reviewable UI slice that can merge independently of
the backend reliability, database backup, workflow, and board QoL PRs.

## What Changed

- Virtualized long issue chat threads and stabilized
anchor/jump-to-latest behavior for large histories.
- Added incremental issue-list row loading and tests for
scroll-triggered pagination behavior.
- Hardened markdown body rendering and markdown editor behavior around
HTML tags, image drops, code-copy UI, and escaped newline handling.
- Added a long-thread measurement harness at
`scripts/measure-issue-chat-long-thread.mjs` plus
`perf:issue-chat-long-thread`.
- Added focused UI/lib regression coverage for thread rendering,
markdown, optimistic comments, and message building.

## Verification

- `pnpm install --frozen-lockfile`
- `pnpm exec vitest run ui/src/components/IssueChatThread.test.tsx
ui/src/components/IssuesList.test.tsx
ui/src/components/MarkdownBody.test.tsx
ui/src/components/MarkdownEditor.test.tsx
ui/src/lib/issue-chat-messages.test.ts
ui/src/lib/optimistic-issue-comments.test.ts`
- Result: 6 test files passed, 170 tests passed.
- UI screenshots not included because this PR is covered by targeted
component tests and does not introduce a new page layout.

## Risks

- Virtualization changes can affect scroll anchoring in edge cases on
very long threads.
- Markdown/editor hardening changes are intentionally defensive, but
malformed content may render differently than before.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex, GPT-5.5, code execution and GitHub CLI tool use, medium
reasoning effort.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-04-30 13:18:01 -05:00
403 changed files with 69358 additions and 2524 deletions

View File

@@ -14,7 +14,7 @@ permissions:
jobs:
build-and-push:
runs-on: ubuntu-latest
timeout-minutes: 30
timeout-minutes: 60
concurrency:
group: docker-${{ github.ref }}
cancel-in-progress: true

View File

@@ -23,7 +23,9 @@ jobs:
- name: Block manual lockfile edits
if: github.head_ref != 'chore/refresh-lockfile'
run: |
changed="$(git diff --name-only "${{ github.event.pull_request.base.sha }}" "${{ github.event.pull_request.head.sha }}")"
# Diff the PR branch against its merge base so recent base-branch commits
# do not masquerade as changes made by the PR itself.
changed="$(git diff --name-only "${{ github.event.pull_request.base.sha }}...${{ github.event.pull_request.head.sha }}")"
if printf '%s\n' "$changed" | grep -qx 'pnpm-lock.yaml'; then
echo "Do not commit pnpm-lock.yaml in pull requests. CI owns lockfile updates."
exit 1
@@ -43,9 +45,18 @@ jobs:
- name: Validate Dockerfile deps stage
run: node ./scripts/check-docker-deps-stage.mjs
- name: Validate release package manifest
run: node ./scripts/release-package-map.mjs check
- name: Verify release package bootstrap for changed manifests
run: |
mapfile -t changed_paths < <(git diff --name-only "${{ github.event.pull_request.base.sha }}...${{ github.event.pull_request.head.sha }}")
PAPERCLIP_RELEASE_BOOTSTRAP_BASE_SHA="${{ github.event.pull_request.base.sha }}" \
node ./scripts/check-release-package-bootstrap.mjs "${changed_paths[@]}"
- name: Validate dependency resolution when manifests change
run: |
changed="$(git diff --name-only "${{ github.event.pull_request.base.sha }}" "${{ github.event.pull_request.head.sha }}")"
changed="$(git diff --name-only "${{ github.event.pull_request.base.sha }}...${{ github.event.pull_request.head.sha }}")"
manifest_pattern='(^|/)package\.json$|^pnpm-workspace\.yaml$|^\.npmrc$|^pnpmfile\.(cjs|js|mjs)$'
if printf '%s\n' "$changed" | grep -Eq "$manifest_pattern"; then
pnpm install --lockfile-only --ignore-scripts --no-frozen-lockfile
@@ -74,11 +85,11 @@ jobs:
- name: Install dependencies
run: pnpm install --frozen-lockfile
- name: Typecheck
run: pnpm -r typecheck
- name: Typecheck workspaces whose build scripts skip TypeScript
run: pnpm run typecheck:build-gaps
- name: Run tests
run: pnpm test:run
- name: Run general test suites
run: pnpm test:run:general
- name: Verify release registry test coverage
run: pnpm run test:release-registry
@@ -86,7 +97,76 @@ jobs:
- name: Build
run: pnpm build
- name: Release canary dry run
verify_serialized_server:
name: Verify serialized server suites (${{ matrix.shard_label }})
needs: [policy]
runs-on: ubuntu-latest
timeout-minutes: 20
strategy:
fail-fast: false
matrix:
include:
- shard_index: 0
shard_count: 4
shard_label: 1/4
- shard_index: 1
shard_count: 4
shard_label: 2/4
- shard_index: 2
shard_count: 4
shard_label: 3/4
- shard_index: 3
shard_count: 4
shard_label: 4/4
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Setup pnpm
uses: pnpm/action-setup@v4
with:
version: 9.15.4
- name: Setup Node.js
uses: actions/setup-node@v4
with:
node-version: 24
cache: pnpm
- name: Install dependencies
run: pnpm install --frozen-lockfile
- name: Run serialized server test shard
run: pnpm test:run:serialized -- --shard-index ${{ matrix.shard_index }} --shard-count ${{ matrix.shard_count }}
canary_dry_run:
name: Canary Dry Run
needs: [policy]
runs-on: ubuntu-latest
timeout-minutes: 20
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Setup pnpm
uses: pnpm/action-setup@v4
with:
version: 9.15.4
- name: Setup Node.js
uses: actions/setup-node@v4
with:
node-version: 24
cache: pnpm
- name: Install dependencies
run: pnpm install --frozen-lockfile
# `release.sh` always executes its Step 2/7 workspace build, even when
# `--skip-verify` bypasses the initial verification gate.
- name: Release canary dry run via release.sh internal build
run: |
git checkout -B master HEAD
git checkout -- pnpm-lock.yaml
@@ -115,9 +195,6 @@ jobs:
- name: Install dependencies
run: pnpm install --frozen-lockfile
- name: Build
run: pnpm build
- name: Install Playwright
run: npx playwright install --with-deps chromium

View File

@@ -50,6 +50,9 @@ jobs:
node-version: 24
cache: pnpm
- name: Validate release package manifest
run: node ./scripts/release-package-map.mjs check
- name: Install dependencies
run: pnpm install --no-frozen-lockfile
@@ -89,6 +92,9 @@ jobs:
node-version: 24
cache: pnpm
- name: Validate release package manifest
run: node ./scripts/release-package-map.mjs check
- name: Install dependencies
run: pnpm install --no-frozen-lockfile
@@ -139,6 +145,9 @@ jobs:
node-version: 24
cache: pnpm
- name: Validate release package manifest
run: node ./scripts/release-package-map.mjs check
- name: Install dependencies
run: pnpm install --no-frozen-lockfile
@@ -177,6 +186,9 @@ jobs:
node-version: 24
cache: pnpm
- name: Validate release package manifest
run: node ./scripts/release-package-map.mjs check
- name: Install dependencies
run: pnpm install --no-frozen-lockfile

View File

@@ -22,6 +22,7 @@ COPY packages/shared/package.json packages/shared/
COPY packages/db/package.json packages/db/
COPY packages/adapter-utils/package.json packages/adapter-utils/
COPY packages/mcp-server/package.json packages/mcp-server/
COPY packages/adapters/acpx-local/package.json packages/adapters/acpx-local/
COPY packages/adapters/claude-local/package.json packages/adapters/claude-local/
COPY packages/adapters/codex-local/package.json packages/adapters/codex-local/
COPY packages/adapters/cursor-local/package.json packages/adapters/cursor-local/

View File

@@ -37,6 +37,7 @@
},
"dependencies": {
"@clack/prompts": "^0.10.0",
"@paperclipai/adapter-acpx-local": "workspace:*",
"@paperclipai/adapter-claude-local": "workspace:*",
"@paperclipai/adapter-codex-local": "workspace:*",
"@paperclipai/adapter-cursor-local": "workspace:*",

View File

@@ -1,4 +1,5 @@
import type { CLIAdapterModule } from "@paperclipai/adapter-utils";
import { printAcpxStreamEvent } from "@paperclipai/adapter-acpx-local/cli";
import { printClaudeStreamEvent } from "@paperclipai/adapter-claude-local/cli";
import { printCodexStreamEvent } from "@paperclipai/adapter-codex-local/cli";
import { printCursorStreamEvent } from "@paperclipai/adapter-cursor-local/cli";
@@ -14,6 +15,11 @@ const claudeLocalCLIAdapter: CLIAdapterModule = {
formatStdoutEvent: printClaudeStreamEvent,
};
const acpxLocalCLIAdapter: CLIAdapterModule = {
type: "acpx_local",
formatStdoutEvent: printAcpxStreamEvent,
};
const codexLocalCLIAdapter: CLIAdapterModule = {
type: "codex_local",
formatStdoutEvent: printCodexStreamEvent,
@@ -46,6 +52,7 @@ const openclawGatewayCLIAdapter: CLIAdapterModule = {
const adaptersByType = new Map<string, CLIAdapterModule>(
[
acpxLocalCLIAdapter,
claudeLocalCLIAdapter,
codexLocalCLIAdapter,
openCodeLocalCLIAdapter,

View File

@@ -176,6 +176,58 @@ That means:
See [doc/RELEASE-AUTOMATION-SETUP.md](RELEASE-AUTOMATION-SETUP.md) for the GitHub/npm setup steps.
## Release enrollment for new public packages
Paperclip does not auto-publish every non-private workspace package anymore.
CI publishing is controlled by [`scripts/release-package-manifest.json`](../scripts/release-package-manifest.json).
When you add a new public package:
1. add it to the manifest and decide whether CI should publish it immediately
2. if CI should publish it, bootstrap the package on npm before merge
3. if CI should not publish it yet, keep `"publishFromCi": false`
4. only enable `"publishFromCi": true` after npm trusted publishing is configured for that package
PR CI now checks changed release-enabled package manifests against npm. That catches a missing first-publish bootstrap before the change reaches `master`.
### One-time bootstrap sequence for a new package
The first publish of a brand-new package still needs one human maintainer with npm write access.
After that, trusted publishing can take over.
Example for `@paperclipai/adapter-acpx-local` from the repo root:
```bash
# safe preview
pnpm run release:bootstrap-package -- @paperclipai/adapter-acpx-local
# one-time first publish from an authenticated maintainer machine
pnpm run release:bootstrap-package -- @paperclipai/adapter-acpx-local --publish --otp 123456
```
The helper script:
- checks that the package does not already exist on npm
- builds the target package unless `--skip-build` is passed
- runs `npm pack --dry-run` in the package directory
- only runs the real `npm publish --access public` when `--publish --otp <code>` is provided
For the real `--publish` step, the maintainer machine must already be authenticated to npm.
If `npm whoami` returns `401`, first run `npm logout --registry=https://registry.npmjs.org/` to clear any stale local auth, then run `npm login` or `npm adduser` locally as an npm org member, and finally rerun the helper.
That local human auth is fine for the one-time bootstrap publish; we just do not want the same auth model inside CI.
The helper now requires `--otp <code>` up front for `--publish`, so it fails before the real publish attempt if the one-time password is missing.
After that first publish succeeds:
1. open `https://www.npmjs.com/package/@paperclipai/adapter-acpx-local`
2. go to `Settings``Trusted publishing`
3. add repository `paperclipai/paperclip`
4. set workflow filename to `release.yml`
5. optionally go to `Settings``Publishing access` and enable `Require two-factor authentication and disallow tokens`
6. keep `publishFromCi: true` in [`scripts/release-package-manifest.json`](../scripts/release-package-manifest.json)
Once those steps are done, future canary and stable publishes for that package are automated through GitHub OIDC. The manual step is only the first package creation on npm.
## Rollback model
Rollback does not unpublish anything.

View File

@@ -67,6 +67,27 @@ Why:
- the single `release.yml` workflow handles both canary and stable publishing
- GitHub environments `npm-canary` and `npm-stable` still enforce different approval rules on the GitHub side
### 2.2.1. Newly added public packages need a bootstrap phase
Trusted publishing is configured on the npm package itself, not at the repo scope.
That means a brand-new public package must not be auto-enrolled into CI publishing until its npm package exists and its trusted publisher has been configured.
Repo policy:
1. add every non-private package to [`scripts/release-package-manifest.json`](../scripts/release-package-manifest.json)
2. set `"publishFromCi": true` only when CI is expected to publish that package
3. if the package is not ready for CI publishing yet, keep `"publishFromCi": false`
4. complete the package bootstrap before merging any PR that changes a release-enabled new package
Bootstrap sequence for a new package:
1. publish the package once from a trusted maintainer machine using normal npm auth
2. open that package on npm and add the `paperclipai/paperclip` trusted publisher for `.github/workflows/release.yml`
3. rerun or dry-run the release flow as needed to confirm CI publishing now works
4. only then enable `"publishFromCi": true`
PR CI enforces this by checking changed release-enabled package manifests against npm. That keeps `master` canary publishing healthy while preserving the no-long-lived-token model for normal CI releases.
### 2.3. Verify trusted publishing before removing old auth
After the workflows are live:

View File

@@ -150,7 +150,7 @@ Invariant: every business record belongs to exactly one company.
- `capabilities` text null
- `adapter_type` text; built-ins include `process`, `http`, `claude_local`, `codex_local`, `gemini_local`, `opencode_local`, `pi_local`, `cursor`, and `openclaw_gateway`
- `adapter_config` jsonb not null
- `runtime_config` jsonb not null default `{}`
- `runtime_config` jsonb not null default `{}`; may include Paperclip runtime policy such as `modelProfiles.cheap.adapterConfig` for an optional low-cost model lane that does not change the primary adapter config
- `default_environment_id` uuid fk `environments.id` null
- `context_mode` enum: `thin | fat` default `thin`
- `budget_monthly_cents` int not null default 0
@@ -676,7 +676,7 @@ Per-agent schedule fields in `adapter_config`:
- `enabled` boolean
- `intervalSec` integer (minimum 30)
- `maxConcurrentRuns` integer; new agents default to `5`
- `maxConcurrentRuns` integer; new agents default to `20`; scheduler clamps configured values to `1..50`
Scheduler must skip invocation when:

View File

@@ -67,13 +67,15 @@ This is the right state for:
- waiting on another issue
- waiting on a human decision
- waiting on an external dependency or system
- waiting on an external dependency or system when Paperclip does not own a scheduled re-check
- work that automatic recovery could not safely continue
### `in_review`
Execution work is paused because the next move belongs to a reviewer or approver, not the current executor.
An external review service can also be a valid review path when the issue keeps an agent assignee and has an active one-shot monitor that will wake that assignee to check the service later.
### `done`
The work is complete and terminal.
@@ -164,6 +166,7 @@ The valid action-path primitives are:
- a queued wake or continuation that can be delivered to the responsible agent
- a typed execution-policy participant, such as `executionState.currentParticipant`
- a pending issue-thread interaction or linked approval that is waiting for a specific responder
- a one-shot issue monitor (`executionPolicy.monitor.nextCheckAt`) that will wake the assignee for a future check
- a human owner via `assigneeUserId`
- a first-class blocker chain whose unresolved leaf issues are themselves healthy
- an open explicit recovery issue that names the owner and action needed to restore liveness
@@ -188,6 +191,7 @@ A healthy active-work state means at least one of these is true:
- there is an active run for the issue
- there is already a queued continuation wake
- there is an active one-shot monitor that will wake the assignee for a future check
- there is an open explicit recovery issue for the lost execution path
An agent-owned `in_progress` issue is stalled when it has no active run, no queued continuation, and no explicit recovery surface. A still-running but silent process is not automatically stalled; it is handled by the active-run watchdog contract.
@@ -202,11 +206,34 @@ A healthy `in_review` issue has at least one valid action path:
- a pending issue-thread interaction or linked approval waiting for a named responder
- a human owner via `assigneeUserId`
- an active run or queued wake that is expected to process the review state
- an active one-shot monitor for an external service or async review loop that the assignee owns
- an open explicit recovery issue for an ambiguous review handoff
Agent-assigned `in_review` with no typed participant is only healthy when one of the other paths exists. Assignment to the same agent that produced the handoff is not, by itself, a review path.
An `in_review` issue is stalled when it has no typed participant, no pending interaction or approval, no user owner, no active run, no queued wake, and no explicit recovery issue. Paperclip should surface that state as recovery work rather than silently completing the issue or leaving blocker chains parked indefinitely.
An `in_review` issue is stalled when it has no typed participant, no pending interaction or approval, no user owner, no active monitor, no active run, no queued wake, and no explicit recovery issue. Paperclip should surface that state as recovery work rather than silently completing the issue or leaving blocker chains parked indefinitely.
### Issue monitors
An issue monitor is a one-shot deferred action path for agent-owned issues in `in_progress` or `in_review`.
Use a monitor when the current assignee owns a future check against an async system or external service. Examples include Greptile review loops, GitHub checks, Vercel deployments, or provider jobs where the agent should come back later and decide what happens next.
Monitor policy lives under `executionPolicy.monitor` and includes:
- `nextCheckAt`: when Paperclip should wake the assignee
- `notes`: non-secret instructions for what the assignee should check
- `serviceName`: optional non-secret external-service context
- `externalRef`: optional external-service reference input; Paperclip treats it as secret-adjacent, redacts it before persistence/visibility, and omits it from activity and wake payloads
- `timeoutAt`, `maxAttempts`, and `recoveryPolicy`: optional recovery hints for bounded waits
Monitors are not recurring intervals. When a monitor fires, Paperclip clears the scheduled monitor and queues an `issue_monitor_due` wake for the assignee. If the external service is still pending, the assignee must explicitly re-arm the monitor with a new `nextCheckAt`. If the issue moves to `done`, `cancelled`, an invalid status, or a human/unassigned owner, the monitor is cleared.
Because `serviceName` and `notes` remain visible in issue activity and wake context, operators should keep them short and non-secret. Put enough context for the assignee to know what to inspect, but do not include signed URLs, bearer tokens, customer secrets, tenant-private identifiers, or provider links with embedded credentials.
Monitor bounds are enforced. Paperclip rejects attempts to re-arm a monitor whose `timeoutAt` or `maxAttempts` is already exhausted. When a scheduled monitor reaches an exhausted bound at trigger time, Paperclip clears it and follows `recoveryPolicy`: `wake_owner` queues a bounded recovery wake for the assignee, `create_recovery_issue` opens visible recovery work, and `escalate_to_board` records a board-visible escalation comment/activity.
Use `blocked` instead of a monitor when no Paperclip assignee owns a responsible polling path. In that case, name the external owner/action or create first-class recovery/blocker work.
### `blocked`

View File

@@ -13,7 +13,9 @@ It is intentionally narrower than [PLUGIN_SPEC.md](./PLUGIN_SPEC.md). The spec i
- Plugin database migrations are restricted to a host-derived plugin namespace.
- Plugin-owned JSON API routes must be declared in the manifest and are mounted
only under `/api/plugins/:pluginId/api/*`.
- There is no host-provided shared React component kit for plugins yet.
- The host provides a small shared React component kit through
`@paperclipai/plugin-sdk/ui`; use it for common Paperclip controls before
building custom versions.
- `ctx.assets` is not supported in the current runtime.
## Scaffold a plugin
@@ -168,6 +170,187 @@ Mount surfaces currently wired in the host include:
- `commentAnnotation`
- `commentContextMenuItem`
## Shared host components
Use shared components from `@paperclipai/plugin-sdk/ui` when the plugin needs a
Paperclip-native control. The host owns the implementation, so plugins inherit
the board's current styling, ordering, recent selections, and dark-mode behavior
without importing `ui/src` internals.
Currently exposed components include:
- `MarkdownBlock` and `MarkdownEditor` for rendered and editable markdown.
- `FileTree` for serializable file and directory trees.
- `IssuesList` for a native company-scoped issue table.
- `AssigneePicker` for the same agent/user selector used in the new issue pane.
Use the controlled `value` format `agent:<id>`, `user:<id>`, or `""`.
- `ProjectPicker` for the same project selector used in the new issue pane.
Use the controlled project id value, or `""` for no project.
- `ManagedRoutinesList` for plugin-owned routine settings pages.
```tsx
import { AssigneePicker, ProjectPicker } from "@paperclipai/plugin-sdk/ui";
export function PluginAssignmentControls({ companyId }: { companyId: string }) {
const [assignee, setAssignee] = useState("");
const [projectId, setProjectId] = useState("");
return (
<>
<AssigneePicker
companyId={companyId}
value={assignee}
onChange={(value) => setAssignee(value)}
/>
<ProjectPicker
companyId={companyId}
value={projectId}
onChange={setProjectId}
/>
</>
);
}
```
## File and path UI
Plugin UI often needs to render a file tree, accept a folder path, or browse a
project workspace. There are three different surfaces for that, and they map to
different trust and data-flow boundaries. Pick the surface that matches the
data the plugin actually has.
### When to use the shared `FileTree`
Use `FileTree` from `@paperclipai/plugin-sdk/ui` whenever the plugin only needs
to render a serializable file/directory list and react to selection or
expand/collapse. The host owns the implementation, so plugin UI inherits the
board's icons, indent, focus ring, and dark-mode styling without importing host
internals.
```tsx
import {
FileTree,
type FileTreeNode,
} from "@paperclipai/plugin-sdk/ui";
const nodes: FileTreeNode[] = [
{ name: "AGENTS.md", path: "AGENTS.md", kind: "file", children: [] },
{
name: "wiki",
path: "wiki",
kind: "dir",
children: [
{ name: "index.md", path: "wiki/index.md", kind: "file", children: [] },
],
},
];
export function WikiTree() {
const [expanded, setExpanded] = useState<Set<string>>(() => new Set(["wiki"]));
const [selected, setSelected] = useState<string | null>(null);
return (
<FileTree
nodes={nodes}
selectedFile={selected}
expandedPaths={expanded}
onSelectFile={(path) => setSelected(path)}
onToggleDir={(path) =>
setExpanded((current) => {
const next = new Set(current);
next.has(path) ? next.delete(path) : next.add(path);
return next;
})
}
/>
);
}
```
Good fits:
- LLM Wiki page navigation in `packages/plugins/plugin-llm-wiki` builds a
`FileTreeNode[]` from worker query results and renders it through `FileTree`.
- The example `plugin-file-browser-example` lazily fetches a directory's
children through a `loadFileList` action when `onToggleDir` fires, then
merges the children into the local tree state — letting the shared component
handle rendering and selection.
Boundary rules:
- Keep the prop surface serializable (`nodes`, `expandedPaths`, `checkedPaths`,
`fileBadges`, `fileTones`). Do not pass arbitrary render functions across the
plugin/host boundary in v1; the supported escape hatches are
`fileBadges` (status pill keyed by path) and `fileTones` (row tone keyed by
path).
- Do not import the host's `FileTree.tsx` or any `ui/src/*` module. The SDK
declaration is the only supported import path for plugin UI.
- The shared `FileTree` is for rendering and selection. Plugin-specific editors,
ingest flows, query forms, and lint runs stay inside the plugin and do not
belong as `FileTree` props.
### When to declare `localFolders`
When the plugin needs operator-configured filesystem roots — typically for
trusted local plugins like wiki tooling — declare `localFolders[]` on the
manifest and add the `local.folders` capability. The host renders a settings
surface for the operator to set the absolute path, validates the path
server-side (containment, symlinks, required files/directories), and exposes
`ctx.localFolders.readText()` and `ctx.localFolders.writeTextAtomic()` in the
worker.
```ts
export const manifest = {
capabilities: ["local.folders"],
localFolders: [
{
folderKey: "content-root",
displayName: "Content root",
access: "readWrite",
requiredDirectories: ["sources", "pages"],
requiredFiles: ["schema.md"],
},
],
};
```
Use this when:
- The data lives outside any project workspace.
- Reads and writes need company-scoped configuration.
- The operator picks the path once in plugin settings and the worker resolves
files relative to that root.
Do not use `localFolders` to grant the UI direct browser-side access to the
filesystem — there is no such capability. The browser still goes through the
worker via `getData` / `performAction`, and the worker only exposes paths it
chose to expose.
### When to keep worker-mediated project workspace browsing
When the data lives inside an existing project workspace, keep the browsing
flow worker-mediated:
- The worker uses `ctx.projects.listWorkspaces()` to resolve the workspace
path, then reads its filesystem with normal Node APIs.
- The plugin UI calls a `getData` handler for the root listing and an action
for lazy children, then renders them through `FileTree`.
- The worker is the only side that touches the disk. The browser receives a
serializable tree and never sees raw absolute paths it can replay.
The example `plugin-file-browser-example` is the reference for this pattern:
the worker registers `fileList` (data) and `loadFileList` (action) over the
same handler, and the UI uses the action for on-toggle directory loading so the
shared `FileTree` stays the rendering surface.
### Mixing surfaces
A single plugin can use more than one of these. The LLM Wiki uses
`localFolders` for its content root, then renders the resulting page list
through `FileTree`. The file browser example uses `ctx.projects.listWorkspaces`
to pick a workspace and renders its on-disk tree through `FileTree` with lazy
loading. Pick the boundary per data source, not per plugin.
## Company routes
Plugins may declare a `page` slot with `routePath` to own a company route like:

View File

@@ -27,7 +27,7 @@ Current limitations to keep in mind:
- Published npm packages are the intended install artifact for deployed plugins.
- The repo example plugins under `packages/plugins/examples/` are development conveniences. They work from a source checkout and should not be assumed to exist in a generic published build unless they are explicitly shipped with that build.
- Dynamic plugin install is not yet cloud-ready for horizontally scaled or ephemeral deployments. There is no shared artifact store, install coordination, or cross-node distribution layer yet.
- The current runtime does not yet ship a real host-provided plugin UI component kit, and it does not support plugin asset uploads/reads. Treat those as future-scope ideas in this spec, not current implementation promises.
- The current runtime ships a small host-provided plugin UI component kit through `@paperclipai/plugin-sdk/ui`, but does not support plugin asset uploads/reads yet. Treat plugin asset APIs as future-scope ideas, not current implementation promises.
- Scoped plugin API routes are JSON-only and must be declared in `apiRoutes`.
They mount under `/api/plugins/:pluginId/api/*`; plugins cannot shadow core
API routes.
@@ -976,13 +976,23 @@ export function DashboardWidget({ context }: PluginWidgetProps) {
The SDK includes a `ui` subpath export that plugin frontends import. This subpath provides:
- **Bridge hooks**: `usePluginData(key, params)`, `usePluginAction(key)`, `useHostContext()`
- **Bridge hooks**: `usePluginData(key, params)`, `usePluginAction(key)`, `useHostContext()`, `useHostNavigation()`
- **Design tokens**: colors, spacing, typography, shadows matching the host theme
- **Shared components**: `MetricCard`, `StatusBadge`, `DataTable`, `LogView`, `ActionBar`, `Spinner`, etc.
- **Type definitions**: `PluginPageProps`, `PluginWidgetProps`, `PluginDetailTabProps`
Plugins are encouraged but not required to use the shared components. A plugin may render entirely custom UI as long as it communicates through the bridge.
`useHostNavigation()` is the supported way for plugin UI to navigate to
Paperclip-internal pages. It exposes `resolveHref(to)`, `navigate(to,
options?)`, and `linkProps(to, options?)`. Plugin links should prefer
`linkProps()` so anchors keep real `href` values for copy-link, modifier-click,
middle-click, and open-in-new-tab behavior while plain left-clicks route through
the host SPA router. The host resolves company-scoped paths against the active
company prefix without double-prefixing already-prefixed paths. Plugin UI should
not use raw same-origin `href`s or `window.location.assign()` for internal
Paperclip navigation because those can force a full document reload.
### 19.0.2 Bundle Isolation
Plugin UI bundles are loaded as standard ES modules, not iframed. This gives plugins full rendering performance and access to the host's design tokens.
@@ -1062,6 +1072,11 @@ The host SDK ships shared components that plugins can import to quickly build UI
| `LogView` | Scrollable log output with timestamps | Webhook deliveries, job output, process logs |
| `JsonTree` | Collapsible JSON tree for debugging | Raw API responses, plugin state inspection |
| `Spinner` | Loading indicator | Data fetch states |
| `FileTree` | Host-styled file/directory tree | Wiki pages, workspace files, import previews |
| `IssuesList` | Host issue list | Plugin pages that need a native issue view |
| `AssigneePicker` | Host assignee picker for agents and board users | Creating issues, assigning routines, filtering work |
| `ProjectPicker` | Host project picker | Creating issues, scoping dashboards, filtering work |
| `ManagedRoutinesList` | Host routine list | Plugin settings pages that manage routines |
Plugins may also use entirely custom components. The shared components exist to reduce boilerplate and keep visual consistency, not to limit what plugins can render.

Binary file not shown.

After

Width:  |  Height:  |  Size: 182 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 108 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 191 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 121 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 183 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 105 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 188 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 106 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 335 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 151 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 74 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 74 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 74 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 74 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 88 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 87 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 41 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 40 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 36 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 36 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 180 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 701 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 316 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 694 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 546 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 701 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 316 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 694 KiB

View File

@@ -15,9 +15,12 @@
"build-storybook": "pnpm --filter @paperclipai/ui build-storybook",
"build": "pnpm run preflight:workspace-links && pnpm -r build",
"typecheck": "pnpm run preflight:workspace-links && pnpm -r typecheck",
"typecheck:build-gaps": "pnpm run preflight:workspace-links && node scripts/run-typecheck-build-gaps.mjs",
"test": "pnpm run test:run",
"test:watch": "pnpm run preflight:workspace-links && vitest",
"test:run": "pnpm run preflight:workspace-links && node scripts/run-vitest-stable.mjs",
"test:run:general": "pnpm run preflight:workspace-links && pnpm --filter @paperclipai/plugin-sdk build && node scripts/run-vitest-stable.mjs --mode general",
"test:run:serialized": "pnpm run preflight:workspace-links && pnpm --filter @paperclipai/plugin-sdk build && node scripts/run-vitest-stable.mjs --mode serialized",
"db:generate": "pnpm --filter @paperclipai/db generate",
"db:migrate": "pnpm --filter @paperclipai/db migrate",
"issue-references:backfill": "pnpm run preflight:workspace-links && tsx scripts/backfill-issue-reference-mentions.ts",
@@ -30,19 +33,22 @@
"release:stable": "./scripts/release.sh stable",
"release:github": "./scripts/create-github-release.sh",
"release:rollback": "./scripts/rollback-latest.sh",
"release:bootstrap-package": "node scripts/bootstrap-npm-package.mjs",
"check:tokens": "node scripts/check-forbidden-tokens.mjs",
"docs:dev": "cd docs && npx mintlify dev",
"smoke:openclaw-join": "./scripts/smoke/openclaw-join.sh",
"smoke:openclaw-docker-ui": "./scripts/smoke/openclaw-docker-ui.sh",
"smoke:openclaw-sse-standalone": "./scripts/smoke/openclaw-sse-standalone.sh",
"test:release-registry": "node --test scripts/verify-release-registry-state.test.mjs",
"smoke:terminal-bench-loop-skill": "node scripts/smoke/terminal-bench-loop-skill-smoke.mjs",
"test:release-registry": "node --test scripts/verify-release-registry-state.test.mjs scripts/release-package-map.test.mjs scripts/check-release-package-bootstrap.test.mjs",
"test:e2e": "npx playwright test --config tests/e2e/playwright.config.ts",
"test:e2e:headed": "npx playwright test --config tests/e2e/playwright.config.ts --headed",
"test:e2e:multiuser-authenticated": "npx playwright test --config tests/e2e/playwright-multiuser-authenticated.config.ts",
"evals:smoke": "cd evals/promptfoo && npx promptfoo@0.103.3 eval",
"test:release-smoke": "npx playwright test --config tests/release-smoke/playwright.config.ts",
"test:release-smoke:headed": "npx playwright test --config tests/release-smoke/playwright.config.ts --headed",
"metrics:paperclip-commits": "tsx scripts/paperclip-commit-metrics.ts"
"metrics:paperclip-commits": "tsx scripts/paperclip-commit-metrics.ts",
"perf:issue-chat-long-thread": "node scripts/measure-issue-chat-long-thread.mjs"
},
"devDependencies": {
"@playwright/test": "^1.58.2",

View File

@@ -55,9 +55,15 @@ describe("command managed runtime", () => {
...process.env,
...input.env,
};
const command = input.command === "sh" ? "/bin/sh" : input.command;
const command =
input.command === "sh" ? "/bin/sh" : input.command === "bash" ? "/bin/bash" : input.command;
const args = [...(input.args ?? [])];
if (input.stdin != null && input.command === "sh" && args[0] === "-lc" && typeof args[1] === "string") {
if (
input.stdin != null &&
(input.command === "sh" || input.command === "bash") &&
args[0] === "-lc" &&
typeof args[1] === "string"
) {
env.PAPERCLIP_TEST_STDIN = input.stdin;
args[1] = `printf '%s' \"$PAPERCLIP_TEST_STDIN\" | (${args[1]})`;
}

View File

@@ -6,6 +6,7 @@ import {
type SandboxManagedRuntimeClient,
type SandboxRemoteExecutionSpec,
} from "./sandbox-managed-runtime.js";
import { preferredShellForSandbox } from "./sandbox-shell.js";
import type { RunProcessResult } from "./server-utils.js";
export interface CommandManagedRuntimeRunner {
@@ -23,10 +24,10 @@ export interface CommandManagedRuntimeRunner {
export interface CommandManagedRuntimeSpec {
providerKey?: string | null;
shellCommand?: "bash" | "sh" | null;
leaseId?: string | null;
remoteCwd: string;
timeoutMs?: number | null;
paperclipApiUrl?: string | null;
}
export type CommandManagedRuntimeAsset = SandboxManagedRuntimeAsset;
@@ -58,10 +59,12 @@ export function createCommandManagedRuntimeClient(input: {
runner: CommandManagedRuntimeRunner;
remoteCwd: string;
timeoutMs: number;
shellCommand?: "bash" | "sh" | null;
}): SandboxManagedRuntimeClient {
const shellCommand = preferredShellForSandbox(input.shellCommand);
const runShell = async (script: string, opts: { stdin?: string; timeoutMs?: number } = {}) => {
const result = await input.runner.execute({
command: "sh",
command: shellCommand,
args: ["-lc", script],
cwd: input.remoteCwd,
stdin: opts.stdin,
@@ -112,7 +115,7 @@ export function createCommandManagedRuntimeClient(input: {
},
remove: async (remotePath) => {
const result = await input.runner.execute({
command: "sh",
command: shellCommand,
args: ["-lc", `rm -rf ${shellQuote(remotePath)}`],
cwd: input.remoteCwd,
timeoutMs: input.timeoutMs,
@@ -121,7 +124,7 @@ export function createCommandManagedRuntimeClient(input: {
},
run: async (command, options) => {
const result = await input.runner.execute({
command: "sh",
command: shellCommand,
args: ["-lc", command],
cwd: input.remoteCwd,
timeoutMs: options.timeoutMs,
@@ -151,17 +154,18 @@ export async function prepareCommandManagedRuntime(input: {
remoteCwd: workspaceRemoteDir,
timeoutMs,
apiKey: null,
paperclipApiUrl: input.spec.paperclipApiUrl ?? null,
};
const client = createCommandManagedRuntimeClient({
runner: input.runner,
remoteCwd: workspaceRemoteDir,
timeoutMs,
shellCommand: input.spec.shellCommand,
});
const shellCommand = preferredShellForSandbox(input.spec.shellCommand);
if (input.installCommand?.trim()) {
const result = await input.runner.execute({
command: "sh",
command: shellCommand,
args: ["-lc", input.installCommand.trim()],
cwd: workspaceRemoteDir,
timeoutMs,

View File

@@ -0,0 +1,21 @@
export const REDACTED_COMMAND_TEXT_VALUE = "***REDACTED***";
const COMMAND_CLI_SECRET_OPTION_RE =
/(\B-{1,2}(?:api[-_]?key|(?:access[-_]?|auth[-_]?)?token|token|authorization|bearer|secret|passwd|password|credential|jwt|private[-_]?key|cookie|connectionstring)(?:\s+|=)(["']?))[^\s"'`]+(\2)/gi;
const COMMAND_ENV_SECRET_ASSIGNMENT_RE =
/(\b[A-Za-z0-9_]*(?:TOKEN|KEY|SECRET|PASSWORD|PASSWD|AUTHORIZATION|JWT)[A-Za-z0-9_]*\s*=\s*)[^\s"'`]+/gi;
const COMMAND_AUTHORIZATION_BEARER_RE = /(\bAuthorization\s*:\s*Bearer\s+)[^\s"'`]+/gi;
const COMMAND_OPENAI_KEY_RE = /\bsk-[A-Za-z0-9_-]{12,}\b/g;
const COMMAND_GITHUB_TOKEN_RE = /\bgh[pousr]_[A-Za-z0-9_]{20,}\b/g;
const COMMAND_JWT_RE =
/\b[A-Za-z0-9_-]{8,}\.[A-Za-z0-9_-]{8,}\.[A-Za-z0-9_-]{8,}(?:\.[A-Za-z0-9_-]{8,})?\b/g;
export function redactCommandText(command: string, redactedValue = REDACTED_COMMAND_TEXT_VALUE): string {
return command
.replace(COMMAND_AUTHORIZATION_BEARER_RE, `$1${redactedValue}`)
.replace(COMMAND_CLI_SECRET_OPTION_RE, `$1${redactedValue}$3`)
.replace(COMMAND_ENV_SECRET_ASSIGNMENT_RE, `$1${redactedValue}`)
.replace(COMMAND_OPENAI_KEY_RE, redactedValue)
.replace(COMMAND_GITHUB_TOKEN_RE, redactedValue)
.replace(COMMAND_JWT_RE, redactedValue);
}

View File

@@ -7,6 +7,7 @@ import { afterEach, describe, expect, it, vi } from "vitest";
import {
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetToRemoteSpec,
adapterExecutionTargetUsesPaperclipBridge,
runAdapterExecutionTargetProcess,
runAdapterExecutionTargetShellCommand,
startAdapterExecutionTargetPaperclipBridge,
@@ -39,7 +40,8 @@ describe("sandbox adapter execution targets", () => {
onSpawn?: (meta: { pid: number; startedAt: string }) => Promise<void>;
}) => {
counter += 1;
return runChildProcess(`sandbox-run-${counter}`, input.command, input.args ?? [], {
const command = input.command === "bash" ? "/bin/bash" : input.command;
return runChildProcess(`sandbox-run-${counter}`, command, input.args ?? [], {
cwd: input.cwd ?? process.cwd(),
env: input.env ?? {},
stdin: input.stdin,
@@ -103,7 +105,6 @@ describe("sandbox adapter execution targets", () => {
environmentId: "env-1",
leaseId: "lease-1",
remoteCwd: "/workspace",
paperclipTransport: "bridge",
});
});
@@ -140,6 +141,68 @@ describe("sandbox adapter execution targets", () => {
}));
});
it("treats SSH targets as bridge-only", () => {
const target = {
kind: "remote" as const,
transport: "ssh" as const,
remoteCwd: "/workspace",
spec: {
host: "ssh.example.test",
port: 22,
username: "paperclip",
remoteWorkspacePath: "/workspace",
remoteCwd: "/workspace",
privateKey: null,
knownHosts: null,
strictHostKeyChecking: true,
},
};
expect(adapterExecutionTargetUsesPaperclipBridge(target)).toBe(true);
expect(adapterExecutionTargetSessionIdentity(target)).toEqual({
transport: "ssh",
host: "ssh.example.test",
port: 22,
username: "paperclip",
remoteCwd: "/workspace",
});
});
it("uses the provider-declared shell for sandbox helper commands", async () => {
const runner = {
execute: vi.fn(async () => ({
exitCode: 0,
signal: null,
timedOut: false,
stdout: "/home/sandbox",
stderr: "",
pid: null,
startedAt: new Date().toISOString(),
})),
};
const target: AdapterSandboxExecutionTarget = {
kind: "remote",
transport: "sandbox",
providerKey: "custom-provider",
shellCommand: "bash",
remoteCwd: "/workspace",
runner,
};
await runAdapterExecutionTargetShellCommand("run-2b", target, 'printf %s "$HOME"', {
cwd: "/local/workspace",
env: {},
timeoutSec: 7,
});
expect(runner.execute).toHaveBeenCalledWith(expect.objectContaining({
command: "bash",
args: ["-lc", 'printf %s "$HOME"'],
cwd: "/workspace",
timeoutMs: 7000,
}));
});
it("starts a localhost Paperclip bridge for sandbox targets in bridge mode", async () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-execution-target-bridge-"));
cleanupDirs.push(rootDir);
@@ -174,7 +237,6 @@ describe("sandbox adapter execution targets", () => {
environmentId: "env-1",
leaseId: "lease-1",
remoteCwd,
paperclipTransport: "bridge",
runner: createLocalSandboxRunner(),
timeoutMs: 30_000,
};
@@ -252,7 +314,6 @@ describe("sandbox adapter execution targets", () => {
environmentId: "env-1",
leaseId: "lease-1",
remoteCwd,
paperclipTransport: "bridge",
runner: createLocalSandboxRunner(),
timeoutMs: 30_000,
};

View File

@@ -2,6 +2,7 @@ import { afterEach, describe, expect, it, vi } from "vitest";
import * as ssh from "./ssh.js";
import {
adapterExecutionTargetUsesManagedHome,
ensureAdapterExecutionTargetRuntimeCommandInstalled,
resolveAdapterExecutionTargetCwd,
runAdapterExecutionTargetShellCommand,
} from "./execution-target.js";
@@ -161,6 +162,80 @@ describe("runAdapterExecutionTargetShellCommand", () => {
});
});
describe("ensureAdapterExecutionTargetRuntimeCommandInstalled", () => {
afterEach(() => {
vi.restoreAllMocks();
});
it("runs install commands for sandbox targets", async () => {
const runner = {
execute: vi.fn(async () => ({
exitCode: 0,
signal: null,
timedOut: false,
stdout: "",
stderr: "",
pid: null,
startedAt: new Date().toISOString(),
})),
};
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId: "run-install",
target: {
kind: "remote",
transport: "sandbox",
providerKey: "e2b",
remoteCwd: "/remote/workspace",
runner,
},
installCommand: "npm install -g @google/gemini-cli",
cwd: "/local/workspace",
env: { PATH: "/usr/bin" },
timeoutSec: 30,
});
expect(runner.execute).toHaveBeenCalledWith(expect.objectContaining({
command: "sh",
args: ["-lc", "npm install -g @google/gemini-cli"],
cwd: "/remote/workspace",
env: { PATH: "/usr/bin" },
timeoutMs: 30_000,
}));
});
it("skips install commands for SSH targets", async () => {
const runSshCommandSpy = vi.spyOn(ssh, "runSshCommand").mockResolvedValue({
stdout: "",
stderr: "",
});
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId: "run-skip",
target: {
kind: "remote",
transport: "ssh",
remoteCwd: "/srv/paperclip/workspace",
spec: {
host: "ssh.example.test",
port: 22,
username: "ssh-user",
remoteCwd: "/srv/paperclip/workspace",
remoteWorkspacePath: "/srv/paperclip/workspace",
privateKey: null,
knownHosts: null,
strictHostKeyChecking: true,
},
},
installCommand: "npm install -g @google/gemini-cli",
cwd: "/tmp/local",
env: {},
});
expect(runSshCommandSpy).not.toHaveBeenCalled();
});
});
describe("resolveAdapterExecutionTargetCwd", () => {
const sshTarget = {
kind: "remote" as const,

View File

@@ -18,7 +18,7 @@ import {
startSandboxCallbackBridgeServer,
startSandboxCallbackBridgeWorker,
} from "./sandbox-callback-bridge.js";
import { parseSshRemoteExecutionSpec, runSshCommand, shellQuote } from "./ssh.js";
import { createSshCommandManagedRuntimeRunner, parseSshRemoteExecutionSpec, runSshCommand, shellQuote } from "./ssh.js";
import {
ensureCommandResolvable,
resolveCommandForLogs,
@@ -26,6 +26,7 @@ import {
type RunProcessResult,
type TerminalResultCleanupOptions,
} from "./server-utils.js";
import { preferredShellForSandbox } from "./sandbox-shell.js";
export interface AdapterLocalExecutionTarget {
kind: "local";
@@ -39,7 +40,6 @@ export interface AdapterSshExecutionTarget {
environmentId?: string | null;
leaseId?: string | null;
remoteCwd: string;
paperclipApiUrl?: string | null;
spec: SshRemoteExecutionSpec;
}
@@ -47,11 +47,10 @@ export interface AdapterSandboxExecutionTarget {
kind: "remote";
transport: "sandbox";
providerKey?: string | null;
shellCommand?: "bash" | "sh" | null;
environmentId?: string | null;
leaseId?: string | null;
remoteCwd: string;
paperclipApiUrl?: string | null;
paperclipTransport?: "direct" | "bridge";
timeoutMs?: number | null;
runner?: CommandManagedRuntimeRunner;
}
@@ -126,13 +125,9 @@ function resolveDefaultPaperclipApiUrl(): string {
return `http://${runtimeHost}:${runtimePort}`;
}
function resolveSandboxPaperclipTransport(
target: Pick<AdapterSandboxExecutionTarget, "paperclipTransport" | "paperclipApiUrl">,
): "direct" | "bridge" {
if (target.paperclipTransport === "direct" || target.paperclipTransport === "bridge") {
return target.paperclipTransport;
}
return target.paperclipApiUrl ? "direct" : "bridge";
function isBridgeDebugEnabled(env: NodeJS.ProcessEnv): boolean {
const value = env.PAPERCLIP_BRIDGE_DEBUG?.trim().toLowerCase();
return value === "1" || value === "true" || value === "yes";
}
function isAdapterExecutionTargetInstance(value: unknown): value is AdapterExecutionTarget {
@@ -180,21 +175,10 @@ export function resolveAdapterExecutionTargetCwd(
return adapterExecutionTargetRemoteCwd(target, localFallbackCwd);
}
export function adapterExecutionTargetPaperclipApiUrl(
target: AdapterExecutionTarget | null | undefined,
): string | null {
if (target?.kind !== "remote") return null;
if (target.transport === "ssh") return target.paperclipApiUrl ?? target.spec.paperclipApiUrl ?? null;
if (resolveSandboxPaperclipTransport(target) === "bridge") return null;
return target.paperclipApiUrl ?? null;
}
export function adapterExecutionTargetUsesPaperclipBridge(
target: AdapterExecutionTarget | null | undefined,
): boolean {
return target?.kind === "remote" &&
target.transport === "sandbox" &&
resolveSandboxPaperclipTransport(target) === "bridge";
return target?.kind === "remote";
}
export function describeAdapterExecutionTarget(
@@ -214,6 +198,33 @@ function requireSandboxRunner(target: AdapterSandboxExecutionTarget): CommandMan
);
}
function preferredSandboxShell(target: AdapterSandboxExecutionTarget): "bash" | "sh" {
return preferredShellForSandbox(target.shellCommand);
}
type AdapterCommandCapableExecutionTarget = AdapterSshExecutionTarget | AdapterSandboxExecutionTarget;
function adapterExecutionTargetCommandRunner(target: AdapterCommandCapableExecutionTarget): CommandManagedRuntimeRunner {
if (target.transport === "ssh") {
return createSshCommandManagedRuntimeRunner({
spec: target.spec,
defaultCwd: target.remoteCwd,
maxBufferBytes: DEFAULT_SANDBOX_CALLBACK_BRIDGE_MAX_BODY_BYTES * 4,
});
}
return requireSandboxRunner(target);
}
function adapterExecutionTargetShellCommand(target: AdapterCommandCapableExecutionTarget): "bash" | "sh" {
return target.transport === "ssh" ? "sh" : preferredSandboxShell(target);
}
function adapterExecutionTargetTimeoutMs(
target: AdapterCommandCapableExecutionTarget,
): number | null | undefined {
return target.transport === "sandbox" ? target.timeoutMs : undefined;
}
export async function ensureAdapterExecutionTargetCommandResolvable(
command: string,
target: AdapterExecutionTarget | null | undefined,
@@ -341,8 +352,9 @@ export async function runAdapterExecutionTargetShellCommand(
}
}
const shellCommand = preferredSandboxShell(target);
return await requireSandboxRunner(target).execute({
command: "sh",
command: shellCommand,
args: ["-lc", command],
cwd: target.remoteCwd,
env: options.env,
@@ -381,6 +393,60 @@ export async function readAdapterExecutionTargetHomeDir(
return homeDir.length > 0 ? homeDir : null;
}
export async function ensureAdapterExecutionTargetRuntimeCommandInstalled(input: {
runId: string;
target: AdapterExecutionTarget | null | undefined;
installCommand?: string | null;
detectCommand?: string | null;
cwd: string;
env: Record<string, string>;
timeoutSec?: number;
graceSec?: number;
onLog?: AdapterExecutionTargetShellOptions["onLog"];
}): Promise<void> {
const installCommand = input.installCommand?.trim();
if (!installCommand || input.target?.kind !== "remote" || input.target.transport !== "sandbox") {
return;
}
const detectCommand = input.detectCommand?.trim();
if (detectCommand) {
const probe = await runAdapterExecutionTargetShellCommand(
input.runId,
input.target,
`command -v ${shellQuote(detectCommand)} >/dev/null 2>&1`,
{
cwd: input.cwd,
env: input.env,
timeoutSec: input.timeoutSec,
graceSec: input.graceSec,
},
);
if (!probe.timedOut && probe.exitCode === 0) {
return;
}
}
const result = await runAdapterExecutionTargetShellCommand(
input.runId,
input.target,
installCommand,
{
cwd: input.cwd,
env: input.env,
timeoutSec: input.timeoutSec,
graceSec: input.graceSec,
onLog: input.onLog,
},
);
if (result.timedOut) {
throw new Error(`Timed out while installing the adapter runtime command via: ${installCommand}`);
}
if ((result.exitCode ?? 0) !== 0) {
throw new Error(`Failed to install the adapter runtime command via: ${installCommand}`);
}
}
export async function ensureAdapterExecutionTargetFile(
runId: string,
target: AdapterExecutionTarget | null | undefined,
@@ -458,15 +524,12 @@ export function adapterExecutionTargetSessionIdentity(
): Record<string, unknown> | null {
if (!target || target.kind === "local") return null;
if (target.transport === "ssh") return buildRemoteExecutionSessionIdentity(target.spec);
const paperclipTransport = resolveSandboxPaperclipTransport(target);
return {
transport: "sandbox",
providerKey: target.providerKey ?? null,
environmentId: target.environmentId ?? null,
leaseId: target.leaseId ?? null,
remoteCwd: target.remoteCwd,
paperclipTransport,
...(paperclipTransport === "direct" && target.paperclipApiUrl ? { paperclipApiUrl: target.paperclipApiUrl } : {}),
};
}
@@ -485,9 +548,7 @@ export function adapterExecutionTargetSessionMatches(
readStringMeta(parsedSaved, "providerKey") === current?.providerKey &&
readStringMeta(parsedSaved, "environmentId") === current?.environmentId &&
readStringMeta(parsedSaved, "leaseId") === current?.leaseId &&
readStringMeta(parsedSaved, "remoteCwd") === current?.remoteCwd &&
readStringMeta(parsedSaved, "paperclipTransport") === (current?.paperclipTransport ?? null) &&
readStringMeta(parsedSaved, "paperclipApiUrl") === (current?.paperclipApiUrl ?? null)
readStringMeta(parsedSaved, "remoteCwd") === current?.remoteCwd
);
}
@@ -512,14 +573,12 @@ export function parseAdapterExecutionTarget(value: unknown): AdapterExecutionTar
environmentId: readStringMeta(parsed, "environmentId"),
leaseId: readStringMeta(parsed, "leaseId"),
remoteCwd: spec.remoteCwd,
paperclipApiUrl: readStringMeta(parsed, "paperclipApiUrl") ?? spec.paperclipApiUrl ?? null,
spec,
};
}
if (kind === "remote" && readStringMeta(parsed, "transport") === "sandbox") {
const remoteCwd = readStringMeta(parsed, "remoteCwd");
const paperclipTransport = readStringMeta(parsed, "paperclipTransport");
if (!remoteCwd) return null;
return {
kind: "remote",
@@ -528,11 +587,6 @@ export function parseAdapterExecutionTarget(value: unknown): AdapterExecutionTar
environmentId: readStringMeta(parsed, "environmentId"),
leaseId: readStringMeta(parsed, "leaseId"),
remoteCwd,
paperclipApiUrl: readStringMeta(parsed, "paperclipApiUrl"),
paperclipTransport:
paperclipTransport === "direct" || paperclipTransport === "bridge"
? paperclipTransport
: undefined,
timeoutMs: typeof parsed.timeoutMs === "number" ? parsed.timeoutMs : null,
};
}
@@ -553,7 +607,6 @@ export function adapterExecutionTargetFromRemoteExecution(
environmentId: metadata.environmentId ?? null,
leaseId: metadata.leaseId ?? null,
remoteCwd: ssh.remoteCwd,
paperclipApiUrl: ssh.paperclipApiUrl ?? null,
spec: ssh,
};
}
@@ -612,10 +665,10 @@ export async function prepareAdapterExecutionTargetRuntime(input: {
runner: requireSandboxRunner(target),
spec: {
providerKey: target.providerKey,
shellCommand: target.shellCommand,
leaseId: target.leaseId,
remoteCwd: target.remoteCwd,
timeoutMs: target.timeoutMs,
paperclipApiUrl: target.paperclipApiUrl,
},
adapterKey: input.adapterKey,
workspaceLocalDir: input.workspaceLocalDir,
@@ -703,7 +756,7 @@ export async function startAdapterExecutionTargetPaperclipBridge(input: {
if (!adapterExecutionTargetUsesPaperclipBridge(input.target)) {
return null;
}
if (!input.target || input.target.kind !== "remote" || input.target.transport !== "sandbox") {
if (!input.target || input.target.kind !== "remote") {
return null;
}
@@ -731,6 +784,8 @@ export async function startAdapterExecutionTargetPaperclipBridge(input: {
process.env.PAPERCLIP_RUNTIME_API_URL?.trim() ||
process.env.PAPERCLIP_API_URL?.trim() ||
resolveDefaultPaperclipApiUrl();
const shellCommand = adapterExecutionTargetShellCommand(target);
const runner = adapterExecutionTargetCommandRunner(target);
await onLog(
"stdout",
@@ -742,15 +797,30 @@ export async function startAdapterExecutionTargetPaperclipBridge(input: {
let worker: Awaited<ReturnType<typeof startSandboxCallbackBridgeWorker>> | null = null;
try {
const client = createCommandManagedSandboxCallbackBridgeQueueClient({
runner: requireSandboxRunner(target),
runner,
remoteCwd: target.remoteCwd,
timeoutMs: target.timeoutMs,
timeoutMs: adapterExecutionTargetTimeoutMs(target),
shellCommand,
});
// PAPERCLIP_BRIDGE_DEBUG opts into verbose stdout logs of every bridge
// proxy request/response. The query string is logged verbatim, so callers
// who pass auth tokens or other sensitive values as query parameters
// should be aware those values appear in the host process's stdout when
// this flag is enabled. Only intended for active debugging in trusted
// environments.
const bridgeDebugEnabled = isBridgeDebugEnabled(process.env);
worker = await startSandboxCallbackBridgeWorker({
client,
queueDir,
maxBodyBytes,
handleRequest: async (request) => {
const method = request.method.trim().toUpperCase() || "GET";
if (bridgeDebugEnabled) {
await onLog(
"stdout",
`[paperclip] Bridge proxy ${method} ${request.path}${request.query ? `?${request.query}` : ""}\n`,
);
}
const headers = new Headers();
for (const [key, value] of Object.entries(request.headers)) {
if (value.trim().length === 0) continue;
@@ -758,13 +828,18 @@ export async function startAdapterExecutionTargetPaperclipBridge(input: {
}
headers.set("authorization", `Bearer ${hostApiToken}`);
headers.set("x-paperclip-run-id", input.runId);
const method = request.method.trim().toUpperCase() || "GET";
const response = await fetch(buildBridgeForwardUrl(hostApiUrl, request), {
method,
headers,
...(method === "GET" || method === "HEAD" ? {} : { body: request.body }),
signal: AbortSignal.timeout(30_000),
});
if (bridgeDebugEnabled) {
await onLog(
"stdout",
`[paperclip] Bridge proxy response ${response.status} for ${method} ${request.path}${request.query ? `?${request.query}` : ""}\n`,
);
}
return {
status: response.status,
headers: buildBridgeResponseHeaders(response),
@@ -773,14 +848,15 @@ export async function startAdapterExecutionTargetPaperclipBridge(input: {
},
});
server = await startSandboxCallbackBridgeServer({
runner: requireSandboxRunner(target),
runner,
remoteCwd: target.remoteCwd,
assetRemoteDir,
queueDir,
bridgeToken,
bridgeAsset,
timeoutMs: target.timeoutMs,
timeoutMs: adapterExecutionTargetTimeoutMs(target),
maxBodyBytes,
shellCommand,
});
} catch (error) {
await Promise.allSettled([

View File

@@ -20,11 +20,14 @@ export type {
AdapterSkillContext,
AdapterSessionCodec,
AdapterModel,
AdapterModelProfileKey,
AdapterModelProfileDefinition,
HireApprovedPayload,
HireApprovedHookResult,
ConfigFieldOption,
ConfigFieldSchema,
AdapterConfigSchema,
AdapterRuntimeCommandSpec,
ServerAdapterModule,
QuotaWindow,
ProviderQuotaResult,
@@ -53,6 +56,10 @@ export {
redactHomePathUserSegmentsInValue,
redactTranscriptEntryPaths,
} from "./log-redaction.js";
export {
REDACTED_COMMAND_TEXT_VALUE,
redactCommandText,
} from "./command-redaction.js";
export { inferOpenAiCompatibleBiller } from "./billing.js";
// Keep the root adapter-utils entry browser-safe because the UI imports it.
// The sandbox callback bridge stays available via its dedicated subpath export.

View File

@@ -44,7 +44,6 @@ export function buildRemoteExecutionSessionIdentity(spec: SshRemoteExecutionSpec
port: spec.port,
username: spec.username,
remoteCwd: spec.remoteCwd,
...(spec.paperclipApiUrl ? { paperclipApiUrl: spec.paperclipApiUrl } : {}),
} as const;
}
@@ -58,8 +57,7 @@ export function remoteExecutionSessionMatches(saved: unknown, current: SshRemote
asString(parsedSaved.host) === currentIdentity.host &&
asNumber(parsedSaved.port) === currentIdentity.port &&
asString(parsedSaved.username) === currentIdentity.username &&
asString(parsedSaved.remoteCwd) === currentIdentity.remoteCwd &&
asString(parsedSaved.paperclipApiUrl) === asString(currentIdentity.paperclipApiUrl)
asString(parsedSaved.remoteCwd) === currentIdentity.remoteCwd
);
}

View File

@@ -37,9 +37,15 @@ describe("sandbox callback bridge", () => {
...process.env,
...input.env,
};
const command = input.command === "sh" ? "/bin/sh" : input.command;
const command =
input.command === "sh" ? "/bin/sh" : input.command === "bash" ? "/bin/bash" : input.command;
const args = [...(input.args ?? [])];
if (input.stdin != null && input.command === "sh" && args[0] === "-lc" && typeof args[1] === "string") {
if (
input.stdin != null &&
(input.command === "sh" || input.command === "bash") &&
args[0] === "-lc" &&
typeof args[1] === "string"
) {
env.PAPERCLIP_TEST_STDIN = input.stdin;
args[1] = `printf '%s' \"$PAPERCLIP_TEST_STDIN\" | (${args[1]})`;
}

View File

@@ -4,6 +4,7 @@ import os from "node:os";
import path from "node:path";
import type { CommandManagedRuntimeRunner } from "./command-managed-runtime.js";
import { preferredShellForSandbox } from "./sandbox-shell.js";
import type { RunProcessResult } from "./server-utils.js";
const DEFAULT_BRIDGE_TOKEN_BYTES = 24;
@@ -133,9 +134,10 @@ async function runShell(
cwd: string,
script: string,
timeoutMs: number,
shellCommand: "bash" | "sh" = "sh",
): Promise<RunProcessResult> {
return await runner.execute({
command: "sh",
command: shellCommand,
args: ["-lc", script],
cwd,
timeoutMs,
@@ -266,10 +268,12 @@ export function createCommandManagedSandboxCallbackBridgeQueueClient(input: {
runner: CommandManagedRuntimeRunner;
remoteCwd: string;
timeoutMs?: number | null;
shellCommand?: "bash" | "sh" | null;
}): SandboxCallbackBridgeQueueClient {
const timeoutMs = normalizeTimeoutMs(input.timeoutMs, DEFAULT_BRIDGE_RESPONSE_TIMEOUT_MS);
const shellCommand = preferredShellForSandbox(input.shellCommand);
const runChecked = async (action: string, script: string) =>
requireSuccessfulResult(action, await runShell(input.runner, input.remoteCwd, script, timeoutMs));
requireSuccessfulResult(action, await runShell(input.runner, input.remoteCwd, script, timeoutMs, shellCommand));
return {
makeDir: async (remotePath) => {
@@ -288,6 +292,7 @@ export function createCommandManagedSandboxCallbackBridgeQueueClient(input: {
"fi",
].join("\n"),
timeoutMs,
shellCommand,
);
requireSuccessfulResult(`list ${remotePath}`, result);
return result.stdout
@@ -525,10 +530,12 @@ export async function startSandboxCallbackBridgeServer(input: {
responseTimeoutMs?: number | null;
timeoutMs?: number | null;
nodeCommand?: string;
shellCommand?: "bash" | "sh" | null;
maxQueueDepth?: number | null;
maxBodyBytes?: number | null;
}): Promise<StartedSandboxCallbackBridgeServer> {
const timeoutMs = normalizeTimeoutMs(input.timeoutMs, DEFAULT_BRIDGE_RESPONSE_TIMEOUT_MS);
const shellCommand = preferredShellForSandbox(input.shellCommand);
const directories = sandboxCallbackBridgeDirectories(input.queueDir);
const remoteEntrypoint = path.posix.join(input.assetRemoteDir, SANDBOX_CALLBACK_BRIDGE_ENTRYPOINT);
if (input.bridgeAsset) {
@@ -536,6 +543,7 @@ export async function startSandboxCallbackBridgeServer(input: {
runner: input.runner,
remoteCwd: input.remoteCwd,
timeoutMs,
shellCommand,
});
await assetClient.makeDir(input.assetRemoteDir);
const entrypointSource = await fs.readFile(input.bridgeAsset.entrypoint, "utf8");
@@ -553,7 +561,7 @@ export async function startSandboxCallbackBridgeServer(input: {
});
const nodeCommand = input.nodeCommand?.trim() || "node";
const startResult = await input.runner.execute({
command: "sh",
command: shellCommand,
args: [
"-lc",
[
@@ -594,6 +602,7 @@ export async function startSandboxCallbackBridgeServer(input: {
"exit 1",
].join("\n"),
timeoutMs,
shellCommand,
);
requireSuccessfulResult("wait for sandbox callback bridge readiness", readyResult);
@@ -626,7 +635,7 @@ export async function startSandboxCallbackBridgeServer(input: {
directories,
stop: async () => {
const stopResult = await input.runner.execute({
command: "sh",
command: shellCommand,
args: [
"-lc",
[

View File

@@ -13,7 +13,6 @@ export interface SandboxRemoteExecutionSpec {
remoteCwd: string;
timeoutMs: number;
apiKey: string | null;
paperclipApiUrl?: string | null;
}
export interface SandboxManagedRuntimeAsset {
@@ -85,7 +84,6 @@ export function parseSandboxRemoteExecutionSpec(value: unknown): SandboxRemoteEx
remoteCwd,
timeoutMs,
apiKey: asString(parsed.apiKey).trim() || null,
paperclipApiUrl: asString(parsed.paperclipApiUrl).trim() || null,
};
}
@@ -96,7 +94,6 @@ export function buildSandboxExecutionSessionIdentity(spec: SandboxRemoteExecutio
provider: spec.provider,
sandboxId: spec.sandboxId,
remoteCwd: spec.remoteCwd,
...(spec.paperclipApiUrl ? { paperclipApiUrl: spec.paperclipApiUrl } : {}),
} as const;
}
@@ -108,8 +105,7 @@ export function sandboxExecutionSessionMatches(saved: unknown, current: SandboxR
asString(parsedSaved.transport) === currentIdentity.transport &&
asString(parsedSaved.provider) === currentIdentity.provider &&
asString(parsedSaved.sandboxId) === currentIdentity.sandboxId &&
asString(parsedSaved.remoteCwd) === currentIdentity.remoteCwd &&
asString(parsedSaved.paperclipApiUrl) === asString(currentIdentity.paperclipApiUrl)
asString(parsedSaved.remoteCwd) === currentIdentity.remoteCwd
);
}

View File

@@ -0,0 +1,3 @@
export function preferredShellForSandbox(shellCommand: string | null | undefined): "bash" | "sh" {
return shellCommand === "bash" ? "bash" : "sh";
}

View File

@@ -1,12 +1,19 @@
import { randomUUID } from "node:crypto";
import fs from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import { describe, expect, it } from "vitest";
import {
applyPaperclipWorkspaceEnv,
appendWithByteCap,
buildInvocationEnvForLogs,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
materializePaperclipSkillCopy,
renderPaperclipWakePrompt,
runningProcesses,
runChildProcess,
sanitizeSshRemoteEnv,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
} from "./server-utils.js";
@@ -39,6 +46,162 @@ async function waitForTextMatch(read: () => string, pattern: RegExp, timeoutMs =
return read().match(pattern);
}
describe("buildInvocationEnvForLogs", () => {
it("redacts inline secrets from resolved command metadata", () => {
const loggedEnv = buildInvocationEnvForLogs(
{ SAFE_VALUE: "visible" },
{
resolvedCommand: "env OPENAI_API_KEY=sk-live-example custom-acp --token ghp_example_secret",
},
);
expect(loggedEnv.SAFE_VALUE).toBe("visible");
expect(loggedEnv.PAPERCLIP_RESOLVED_COMMAND).toBe(
"env OPENAI_API_KEY=***REDACTED*** custom-acp --token ***REDACTED***",
);
});
});
describe("sanitizeSshRemoteEnv", () => {
it("drops inherited host shell identity variables for SSH remote execution", () => {
expect(
sanitizeSshRemoteEnv(
{
PATH: "/host/bin:/usr/bin",
HOME: "/Users/local",
NVM_DIR: "/Users/local/.nvm",
TMPDIR: "/var/folders/local/T",
XDG_CONFIG_HOME: "/Users/local/.config",
SAFE_VALUE: "visible",
},
{
PATH: "/host/bin:/usr/bin",
HOME: "/Users/local",
NVM_DIR: "/Users/local/.nvm",
TMPDIR: "/var/folders/local/T",
XDG_CONFIG_HOME: "/Users/local/.config",
},
),
).toEqual({
SAFE_VALUE: "visible",
});
});
it("preserves explicit remote overrides even for filtered key names", () => {
expect(
sanitizeSshRemoteEnv(
{
PATH: "/custom/remote/bin:/usr/bin",
HOME: "/home/agent",
TMPDIR: "/tmp",
SAFE_VALUE: "visible",
},
{
PATH: "/host/bin:/usr/bin",
HOME: "/Users/local",
TMPDIR: "/var/folders/local/T",
},
),
).toEqual({
PATH: "/custom/remote/bin:/usr/bin",
HOME: "/home/agent",
TMPDIR: "/tmp",
SAFE_VALUE: "visible",
});
});
it("filters identity keys via case-insensitive match against the inherited env", () => {
expect(
sanitizeSshRemoteEnv(
{
// Caller passed PATH in upper case while the inherited (Windows-style)
// host env exposes it as Path. The lookup must still treat them as
// equal so the leaked host PATH gets stripped.
PATH: "/host/bin:/usr/bin",
HOME: "/host/home",
},
{
Path: "/host/bin:/usr/bin",
home: "/host/home",
},
),
).toEqual({});
});
it("preserves explicitly-set identity keys when the inherited env disagrees in case but not in value", () => {
expect(
sanitizeSshRemoteEnv(
{
PATH: "/explicit/remote/bin",
},
{
Path: "/host/bin:/usr/bin",
},
),
).toEqual({ PATH: "/explicit/remote/bin" });
});
});
describe("materializePaperclipSkillCopy", () => {
it("refuses to materialize into an ancestor of the source", async () => {
const root = await fs.mkdtemp(path.join(os.tmpdir(), "paperclip-skill-copy-"));
try {
const source = path.join(root, "parent", "skill");
await fs.mkdir(source, { recursive: true });
await fs.writeFile(path.join(source, "SKILL.md"), "# skill\n", "utf8");
await expect(materializePaperclipSkillCopy(source, path.join(root, "parent"))).rejects.toThrow(
/ancestor/,
);
await expect(fs.readFile(path.join(source, "SKILL.md"), "utf8")).resolves.toBe("# skill\n");
} finally {
await fs.rm(root, { recursive: true, force: true });
}
});
it("does not delete and recopy an unchanged materialized skill target", async () => {
const root = await fs.mkdtemp(path.join(os.tmpdir(), "paperclip-skill-copy-"));
try {
const source = path.join(root, "source");
const target = path.join(root, "target");
await fs.mkdir(source, { recursive: true });
await fs.writeFile(path.join(source, "SKILL.md"), "# skill\n", "utf8");
const first = await materializePaperclipSkillCopy(source, target);
expect(first.copiedFiles).toBe(1);
await fs.writeFile(path.join(target, "local-marker.txt"), "keep\n", "utf8");
const second = await materializePaperclipSkillCopy(source, target);
expect(second.copiedFiles).toBe(0);
await expect(fs.readFile(path.join(target, "local-marker.txt"), "utf8")).resolves.toBe("keep\n");
} finally {
await fs.rm(root, { recursive: true, force: true });
}
});
it("breaks stale materialization locks left by dead processes", async () => {
const root = await fs.mkdtemp(path.join(os.tmpdir(), "paperclip-skill-copy-"));
try {
const source = path.join(root, "source");
const target = path.join(root, "target");
const lock = `${target}.lock`;
await fs.mkdir(source, { recursive: true });
await fs.writeFile(path.join(source, "SKILL.md"), "# skill\n", "utf8");
await fs.mkdir(lock, { recursive: true });
await fs.writeFile(
path.join(lock, "owner.json"),
JSON.stringify({ pid: 999_999_999, createdAt: "2000-01-01T00:00:00.000Z" }),
"utf8",
);
await expect(materializePaperclipSkillCopy(source, target)).resolves.toMatchObject({ copiedFiles: 1 });
await expect(fs.readFile(path.join(target, "SKILL.md"), "utf8")).resolves.toBe("# skill\n");
} finally {
await fs.rm(root, { recursive: true, force: true });
}
});
});
describe("runChildProcess", () => {
it("does not arm a timeout when timeoutSec is 0", async () => {
const result = await runChildProcess(
@@ -470,6 +633,70 @@ describe("applyPaperclipWorkspaceEnv", () => {
});
});
describe("shapePaperclipWorkspaceEnvForExecution", () => {
it("rewrites workspace env paths for remote execution", () => {
const shaped = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: "/tmp/workspace",
workspaceWorktreePath: "/tmp/worktree",
workspaceHints: [
{
workspaceId: "workspace-1",
cwd: "/tmp/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
},
{
workspaceId: "workspace-2",
cwd: "/tmp/other-workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
},
{
workspaceId: "workspace-3",
repoUrl: "https://github.com/paperclipai/paperclip.git",
},
],
executionTargetIsRemote: true,
executionCwd: "/remote/workspace",
});
expect(shaped).toEqual({
workspaceCwd: "/remote/workspace",
workspaceWorktreePath: null,
workspaceHints: [
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
},
{
workspaceId: "workspace-2",
repoUrl: "https://github.com/paperclipai/paperclip.git",
},
{
workspaceId: "workspace-3",
repoUrl: "https://github.com/paperclipai/paperclip.git",
},
],
});
});
it("leaves local execution workspace paths unchanged", () => {
const workspaceHints = [{ workspaceId: "workspace-1", cwd: "/tmp/workspace" }];
const shaped = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: "/tmp/workspace",
workspaceWorktreePath: "/tmp/worktree",
workspaceHints,
executionTargetIsRemote: false,
executionCwd: "/remote/workspace",
});
expect(shaped).toEqual({
workspaceCwd: "/tmp/workspace",
workspaceWorktreePath: "/tmp/worktree",
workspaceHints,
});
});
});
describe("appendWithByteCap", () => {
it("keeps valid UTF-8 when trimming through multibyte text", () => {
const output = appendWithByteCap("prefix ", "hello — world", 7);

View File

@@ -1,7 +1,9 @@
import { spawn, type ChildProcess } from "node:child_process";
import { createHash, randomUUID } from "node:crypto";
import { constants as fsConstants, promises as fs, type Dirent } from "node:fs";
import path from "node:path";
import { buildSshSpawnTarget, type SshRemoteExecutionSpec } from "./ssh.js";
import { redactCommandText } from "./command-redaction.js";
import type {
AdapterSkillEntry,
AdapterSkillSnapshot,
@@ -76,10 +78,14 @@ export const MAX_CAPTURE_BYTES = 4 * 1024 * 1024;
export const MAX_EXCERPT_BYTES = 32 * 1024;
const TERMINAL_RESULT_SCAN_OVERLAP_CHARS = 64 * 1024;
const SENSITIVE_ENV_KEY = /(key|token|secret|password|passwd|authorization|cookie)/i;
const REDACTED_LOG_VALUE = "***REDACTED***";
const PAPERCLIP_SKILL_ROOT_RELATIVE_CANDIDATES = [
"../../skills",
"../../../../../skills",
];
const MATERIALIZED_SKILL_SENTINEL = ".paperclip-materialized-skill.json";
const MATERIALIZED_SKILL_LOCK_OWNER = "owner.json";
const MATERIALIZED_SKILL_LOCK_STALE_MS = 30_000;
export const DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE = [
"You are agent {{agent.id}} ({{agent.name}}). Continue your Paperclip work.",
@@ -111,6 +117,11 @@ export interface InstalledSkillTarget {
kind: "symlink" | "directory" | "file";
}
export interface MaterializedPaperclipSkillCopyResult {
copiedFiles: number;
skippedSymlinks: string[];
}
interface PersistentSkillSnapshotOptions {
adapterType: string;
availableEntries: PaperclipSkillEntry[];
@@ -780,11 +791,15 @@ export function renderPaperclipWakePrompt(
export function redactEnvForLogs(env: Record<string, string>): Record<string, string> {
const redacted: Record<string, string> = {};
for (const [key, value] of Object.entries(env)) {
redacted[key] = SENSITIVE_ENV_KEY.test(key) ? "***REDACTED***" : value;
redacted[key] = SENSITIVE_ENV_KEY.test(key) ? REDACTED_LOG_VALUE : value;
}
return redacted;
}
export function redactCommandTextForLogs(command: string): string {
return redactCommandText(command, REDACTED_LOG_VALUE);
}
export function buildInvocationEnvForLogs(
env: Record<string, string>,
options: {
@@ -806,7 +821,7 @@ export function buildInvocationEnvForLogs(
const resolvedCommand = options.resolvedCommand?.trim();
if (resolvedCommand) {
merged[options.resolvedCommandEnvKey ?? "PAPERCLIP_RESOLVED_COMMAND"] = resolvedCommand;
merged[options.resolvedCommandEnvKey ?? "PAPERCLIP_RESOLVED_COMMAND"] = redactCommandTextForLogs(resolvedCommand);
}
return redactEnvForLogs(merged);
@@ -870,6 +885,79 @@ export function applyPaperclipWorkspaceEnv(
return env;
}
export function shapePaperclipWorkspaceEnvForExecution(input: {
workspaceCwd?: string | null;
workspaceWorktreePath?: string | null;
workspaceHints?: Array<Record<string, unknown>>;
executionTargetIsRemote?: boolean;
executionCwd?: string | null;
}): {
workspaceCwd: string | null;
workspaceWorktreePath: string | null;
workspaceHints: Array<Record<string, unknown>>;
} {
const workspaceCwd =
typeof input.workspaceCwd === "string" && input.workspaceCwd.trim().length > 0
? input.workspaceCwd.trim()
: null;
const workspaceWorktreePath =
typeof input.workspaceWorktreePath === "string" && input.workspaceWorktreePath.trim().length > 0
? input.workspaceWorktreePath.trim()
: null;
const workspaceHints = Array.isArray(input.workspaceHints) ? input.workspaceHints : [];
if (!input.executionTargetIsRemote) {
return {
workspaceCwd,
workspaceWorktreePath,
workspaceHints,
};
}
const executionCwd =
typeof input.executionCwd === "string" && input.executionCwd.trim().length > 0
? input.executionCwd.trim()
: null;
// On a remote target we must never fall back to the local workspaceCwd —
// doing so leaks host paths into the remote env (the exact failure mode
// this helper exists to prevent). Callers are expected to resolve
// executionCwd via adapterExecutionTargetRemoteCwd before calling this
// helper, which always returns a non-empty string. Surface a warning so
// future callers don't silently regress to the leak.
if (executionCwd === null) {
// eslint-disable-next-line no-console
console.warn(
"[paperclip] shapePaperclipWorkspaceEnvForExecution called with executionCwd=null on a remote target; " +
"stripping workspaceCwd to avoid leaking local paths into the remote environment.",
);
}
const realizedWorkspaceCwd = executionCwd;
const localWorkspaceCwd = workspaceCwd ? path.resolve(workspaceCwd) : null;
const shapedWorkspaceHints = workspaceHints.map((hint) => {
const nextHint = { ...hint };
const hintCwd = typeof nextHint.cwd === "string" ? nextHint.cwd.trim() : "";
if (!hintCwd) return nextHint;
if (localWorkspaceCwd && path.resolve(hintCwd) === localWorkspaceCwd) {
if (realizedWorkspaceCwd) {
nextHint.cwd = realizedWorkspaceCwd;
} else {
delete nextHint.cwd;
}
return nextHint;
}
delete nextHint.cwd;
return nextHint;
});
return {
workspaceCwd: realizedWorkspaceCwd,
workspaceWorktreePath: null,
workspaceHints: shapedWorkspaceHints,
};
}
export function sanitizeInheritedPaperclipEnv(baseEnv: NodeJS.ProcessEnv): NodeJS.ProcessEnv {
const env: NodeJS.ProcessEnv = { ...baseEnv };
for (const key of Object.keys(env)) {
@@ -951,6 +1039,56 @@ function quoteForCmd(arg: string) {
return /[\s"&<>|^()]/.test(escaped) ? `"${escaped}"` : escaped;
}
const SSH_REMOTE_ENV_IDENTITY_KEYS = new Set([
"PATH",
"HOME",
"PWD",
"SHELL",
"USER",
"LOGNAME",
"NVM_DIR",
"TMPDIR",
"TMP",
"TEMP",
"XDG_CONFIG_HOME",
"XDG_CACHE_HOME",
"XDG_DATA_HOME",
"XDG_STATE_HOME",
"XDG_RUNTIME_DIR",
]);
function readEnvValueCaseInsensitive(env: NodeJS.ProcessEnv, key: string): string | undefined {
const direct = env[key];
if (typeof direct === "string") return direct;
const upper = key.toUpperCase();
for (const [candidateKey, candidateValue] of Object.entries(env)) {
if (candidateKey.toUpperCase() === upper && typeof candidateValue === "string") {
return candidateValue;
}
}
return undefined;
}
export function sanitizeSshRemoteEnv(
env: Record<string, string>,
inheritedEnv: NodeJS.ProcessEnv = process.env,
): Record<string, string> {
const sanitized: Record<string, string> = {};
for (const [key, value] of Object.entries(env)) {
const normalizedKey = key.toUpperCase();
if (!SSH_REMOTE_ENV_IDENTITY_KEYS.has(normalizedKey)) {
sanitized[key] = value;
continue;
}
const inheritedValue = readEnvValueCaseInsensitive(inheritedEnv, key);
if (typeof inheritedValue === "string" && inheritedValue === value) {
continue;
}
sanitized[key] = value;
}
return sanitized;
}
function resolveWindowsCmdShell(env: NodeJS.ProcessEnv): string {
const fallbackRoot = env.SystemRoot || process.env.SystemRoot || "C:\\Windows";
return path.join(fallbackRoot, "System32", "cmd.exe");
@@ -976,9 +1114,9 @@ async function resolveSpawnTarget(
spec: remote,
command,
args,
env: Object.fromEntries(
env: sanitizeSshRemoteEnv(Object.fromEntries(
Object.entries(options.remoteEnv ?? {}).filter((entry): entry is [string, string] => typeof entry[1] === "string"),
),
)),
});
return {
command: sshResolved,
@@ -1395,6 +1533,190 @@ export async function ensurePaperclipSkillSymlink(
return "repaired";
}
async function hashSkillDirectory(root: string): Promise<string> {
const hash = createHash("sha256");
async function visit(candidate: string, relativePath: string): Promise<void> {
const stat = await fs.lstat(candidate);
if (stat.isSymbolicLink()) {
hash.update(`symlink:${relativePath}\n`);
return;
}
if (stat.isDirectory()) {
hash.update(`dir:${relativePath}\n`);
const entries = await fs.readdir(candidate, { withFileTypes: true });
entries.sort((left, right) => left.name.localeCompare(right.name));
for (const entry of entries) {
const childRelativePath = relativePath ? `${relativePath}/${entry.name}` : entry.name;
await visit(path.join(candidate, entry.name), childRelativePath);
}
return;
}
if (stat.isFile()) {
hash.update(`file:${relativePath}:${stat.mode}\n`);
hash.update(await fs.readFile(candidate));
hash.update("\n");
return;
}
hash.update(`other:${relativePath}:${stat.mode}\n`);
}
await visit(root, "");
return hash.digest("hex");
}
async function materializedSkillFingerprintMatches(targetRoot: string, sourceFingerprint: string): Promise<boolean> {
try {
const raw = JSON.parse(await fs.readFile(path.join(targetRoot, MATERIALIZED_SKILL_SENTINEL), "utf8")) as unknown;
const parsed = parseObject(raw);
return parsed.version === 1 && parsed.sourceFingerprint === sourceFingerprint;
} catch {
return false;
}
}
async function acquireMaterializeLock(lockDir: string): Promise<() => Promise<void>> {
await fs.mkdir(path.dirname(lockDir), { recursive: true });
const deadline = Date.now() + MATERIALIZED_SKILL_LOCK_STALE_MS;
while (true) {
try {
await fs.mkdir(lockDir);
await fs.writeFile(
path.join(lockDir, MATERIALIZED_SKILL_LOCK_OWNER),
`${JSON.stringify({ pid: process.pid, createdAt: new Date().toISOString() })}\n`,
"utf8",
);
return async () => {
await fs.rm(lockDir, { recursive: true, force: true });
};
} catch (err) {
const code = err && typeof err === "object" ? (err as { code?: unknown }).code : null;
if (code !== "EEXIST") throw err;
if (await removeStaleMaterializeLock(lockDir, MATERIALIZED_SKILL_LOCK_STALE_MS)) continue;
if (Date.now() >= deadline) {
throw new Error(`Timed out waiting for Paperclip skill materialization lock at ${lockDir}`);
}
await new Promise((resolve) => setTimeout(resolve, 50));
}
}
}
function isPidAlive(pid: number): boolean {
if (!Number.isInteger(pid) || pid <= 0) return false;
try {
process.kill(pid, 0);
return true;
} catch (err) {
const code = err && typeof err === "object" ? (err as { code?: unknown }).code : null;
return code === "EPERM";
}
}
async function removeStaleMaterializeLock(lockDir: string, staleMs: number): Promise<boolean> {
const ownerPath = path.join(lockDir, MATERIALIZED_SKILL_LOCK_OWNER);
let shouldRemove = false;
try {
const raw = JSON.parse(await fs.readFile(ownerPath, "utf8")) as unknown;
const owner = parseObject(raw);
const pid = typeof owner.pid === "number" ? owner.pid : 0;
const createdAt = typeof owner.createdAt === "string" ? Date.parse(owner.createdAt) : Number.NaN;
const ageMs = Number.isFinite(createdAt) ? Date.now() - createdAt : staleMs + 1;
shouldRemove = !isPidAlive(pid) || ageMs > staleMs;
} catch {
const stat = await fs.stat(lockDir).catch(() => null);
shouldRemove = !stat || Date.now() - stat.mtimeMs > staleMs;
}
if (!shouldRemove) return false;
await fs.rm(lockDir, { recursive: true, force: true }).catch(() => {});
return true;
}
export async function materializePaperclipSkillCopy(
source: string,
target: string,
): Promise<MaterializedPaperclipSkillCopyResult> {
const sourceRoot = path.resolve(source);
const targetRoot = path.resolve(target);
const relativeTarget = path.relative(sourceRoot, targetRoot);
const relativeSource = path.relative(targetRoot, sourceRoot);
if (
!relativeTarget ||
(!relativeTarget.startsWith("..") && !path.isAbsolute(relativeTarget)) ||
!relativeSource ||
(!relativeSource.startsWith("..") && !path.isAbsolute(relativeSource))
) {
throw new Error("Refusing to materialize a skill into itself, an ancestor, or one of its descendants.");
}
const rootStat = await fs.lstat(sourceRoot);
if (rootStat.isSymbolicLink()) {
throw new Error("Refusing to materialize a skill root that is itself a symlink.");
}
if (!rootStat.isDirectory()) {
throw new Error("Paperclip skills must be directories.");
}
const result: MaterializedPaperclipSkillCopyResult = {
copiedFiles: 0,
skippedSymlinks: [],
};
const lockDir = `${targetRoot}.lock`;
const releaseLock = await acquireMaterializeLock(lockDir);
const tempRoot = `${targetRoot}.tmp-${process.pid}-${randomUUID()}`;
async function copyEntry(sourcePath: string, targetPath: string, relativePath: string): Promise<void> {
const stat = await fs.lstat(sourcePath);
if (stat.isSymbolicLink()) {
result.skippedSymlinks.push(relativePath || ".");
return;
}
if (stat.isDirectory()) {
await fs.mkdir(targetPath, { recursive: true });
const entries = await fs.readdir(sourcePath, { withFileTypes: true });
entries.sort((left, right) => left.name.localeCompare(right.name));
for (const entry of entries) {
const childRelativePath = relativePath ? `${relativePath}/${entry.name}` : entry.name;
await copyEntry(path.join(sourcePath, entry.name), path.join(targetPath, entry.name), childRelativePath);
}
return;
}
if (stat.isFile()) {
await fs.mkdir(path.dirname(targetPath), { recursive: true });
await fs.copyFile(sourcePath, targetPath, fsConstants.COPYFILE_FICLONE).catch(async () => {
await fs.copyFile(sourcePath, targetPath);
});
await fs.chmod(targetPath, stat.mode).catch(() => {});
result.copiedFiles += 1;
}
}
try {
const sourceFingerprint = await hashSkillDirectory(sourceRoot);
if (await materializedSkillFingerprintMatches(targetRoot, sourceFingerprint)) return result;
await copyEntry(sourceRoot, tempRoot, "");
await fs.writeFile(
path.join(tempRoot, MATERIALIZED_SKILL_SENTINEL),
`${JSON.stringify({
version: 1,
sourceFingerprint,
copiedFiles: result.copiedFiles,
skippedSymlinks: result.skippedSymlinks,
}, null, 2)}\n`,
"utf8",
);
if (await materializedSkillFingerprintMatches(targetRoot, sourceFingerprint)) return result;
await fs.rm(targetRoot, { recursive: true, force: true });
await fs.rename(tempRoot, targetRoot);
return result;
} finally {
await fs.rm(tempRoot, { recursive: true, force: true }).catch(() => {});
await releaseLock();
}
}
export async function removeMaintainerOnlySkillSymlinks(
skillsHome: string,
allowedSkillNames: Iterable<string>,

View File

@@ -37,6 +37,7 @@ const ADAPTER_MANAGED_SESSION_POLICY: SessionCompactionPolicy = {
};
export const LEGACY_SESSIONED_ADAPTER_TYPES = new Set([
"acpx_local",
"claude_local",
"codex_local",
"cursor",
@@ -47,6 +48,11 @@ export const LEGACY_SESSIONED_ADAPTER_TYPES = new Set([
]);
export const ADAPTER_SESSION_MANAGEMENT: Record<string, AdapterSessionManagement> = {
acpx_local: {
supportsSessionResume: true,
nativeContextManagement: "confirmed",
defaultSessionCompaction: ADAPTER_MANAGED_SESSION_POLICY,
},
claude_local: {
supportsSessionResume: true,
nativeContextManagement: "confirmed",

View File

@@ -3,6 +3,8 @@ import { constants as fsConstants, createReadStream, createWriteStream, promises
import net from "node:net";
import os from "node:os";
import path from "node:path";
import type { CommandManagedRuntimeRunner } from "./command-managed-runtime.js";
import type { RunProcessResult } from "./server-utils.js";
export interface SshConnectionConfig {
host: string;
@@ -21,7 +23,86 @@ export interface SshCommandResult {
export interface SshRemoteExecutionSpec extends SshConnectionConfig {
remoteCwd: string;
paperclipApiUrl?: string | null;
}
export function createSshCommandManagedRuntimeRunner(input: {
spec: SshRemoteExecutionSpec;
defaultCwd?: string | null;
maxBufferBytes?: number | null;
}): CommandManagedRuntimeRunner {
const defaultCwd = input.defaultCwd?.trim() || input.spec.remoteCwd;
const maxBufferBytes =
typeof input.maxBufferBytes === "number" && Number.isFinite(input.maxBufferBytes) && input.maxBufferBytes > 0
? Math.trunc(input.maxBufferBytes)
: 1024 * 1024;
return {
execute: async (commandInput): Promise<RunProcessResult> => {
const startedAt = new Date().toISOString();
const command = commandInput.command.trim();
const args = commandInput.args ?? [];
const cwd = commandInput.cwd?.trim() || defaultCwd;
const envEntries = Object.entries(commandInput.env ?? {})
.filter((entry): entry is [string, string] => typeof entry[1] === "string");
const envPrefix = envEntries.length > 0
? `env ${envEntries.map(([key, value]) => `${key}=${shellQuote(value)}`).join(" ")} `
: "";
const exportPrefix = envEntries.length > 0
? envEntries.map(([key, value]) => `export ${key}=${shellQuote(value)};`).join(" ") + " "
: "";
const commandScript = command === "sh" || command === "bash"
? args[0] === "-lc" && typeof args[1] === "string"
? `${exportPrefix}${args[1]}`
: `${envPrefix}exec ${[shellQuote(command), ...args.map((arg) => shellQuote(arg))].join(" ")}`
: `${envPrefix}exec ${[shellQuote(command), ...args.map((arg) => shellQuote(arg))].join(" ")}`;
const remoteCommand = `${command === "bash" ? "bash" : "sh"} -lc ${
shellQuote(`cd ${shellQuote(cwd)} && ${commandScript}`)
}`;
try {
const result = await runSshCommand(input.spec, remoteCommand, {
timeoutMs: commandInput.timeoutMs,
maxBuffer: maxBufferBytes,
});
if (result.stdout) await commandInput.onLog?.("stdout", result.stdout);
if (result.stderr) await commandInput.onLog?.("stderr", result.stderr);
return {
exitCode: 0,
signal: null,
timedOut: false,
stdout: result.stdout,
stderr: result.stderr,
pid: null,
startedAt,
};
} catch (error) {
const failure = error as {
stdout?: unknown;
stderr?: unknown;
code?: unknown;
signal?: unknown;
killed?: unknown;
};
const stdout = typeof failure.stdout === "string" ? failure.stdout : "";
const stderr = typeof failure.stderr === "string"
? failure.stderr
: error instanceof Error
? error.message
: String(error);
if (stdout) await commandInput.onLog?.("stdout", stdout);
if (stderr) await commandInput.onLog?.("stderr", stderr);
return {
exitCode: typeof failure.code === "number" ? failure.code : null,
signal: typeof failure.signal === "string" ? failure.signal : null,
timedOut: failure.killed === true,
stdout,
stderr,
pid: null,
startedAt,
};
}
},
};
}
export interface SshEnvLabSupport {
@@ -83,10 +164,6 @@ export function parseSshRemoteExecutionSpec(value: unknown): SshRemoteExecutionS
port: portValue,
username,
remoteCwd,
paperclipApiUrl:
typeof parsed.paperclipApiUrl === "string" && parsed.paperclipApiUrl.trim().length > 0
? parsed.paperclipApiUrl.trim()
: null,
remoteWorkspacePath:
typeof parsed.remoteWorkspacePath === "string" && parsed.remoteWorkspacePath.trim().length > 0
? parsed.remoteWorkspacePath.trim()
@@ -98,50 +175,6 @@ export function parseSshRemoteExecutionSpec(value: unknown): SshRemoteExecutionS
};
}
function normalizeHttpUrlCandidate(value: string): string | null {
const trimmed = value.trim();
if (!trimmed) return null;
try {
const parsed = new URL(trimmed);
if (parsed.protocol !== "http:" && parsed.protocol !== "https:") {
return null;
}
return parsed.origin;
} catch {
return null;
}
}
export async function findReachablePaperclipApiUrlOverSsh(input: {
config: SshConnectionConfig;
candidates: string[];
timeoutMs?: number;
}): Promise<string | null> {
const uniqueCandidates = Array.from(
new Set(
input.candidates
.map((candidate) => normalizeHttpUrlCandidate(candidate))
.filter((candidate): candidate is string => candidate !== null),
),
);
for (const candidate of uniqueCandidates) {
const healthUrl = new URL("/api/health", candidate).toString();
try {
await runSshCommand(
input.config,
`sh -lc ${shellQuote(`curl -fsS -m ${Math.max(1, Math.ceil((input.timeoutMs ?? 5_000) / 1000))} ${shellQuote(healthUrl)} >/dev/null`)}`,
{ timeoutMs: input.timeoutMs ?? 5_000 },
);
return candidate;
} catch {
continue;
}
}
return null;
}
async function execFileText(
file: string,
args: string[],

View File

@@ -125,6 +125,7 @@ export interface AdapterExecutionContext {
runtime: AdapterRuntime;
config: Record<string, unknown>;
context: Record<string, unknown>;
runtimeCommandSpec?: AdapterRuntimeCommandSpec | null;
executionTarget?: AdapterExecutionTarget | null;
/**
* Legacy remote transport view. Prefer `executionTarget`, which is the
@@ -144,6 +145,16 @@ export interface AdapterModel {
label: string;
}
export type AdapterModelProfileKey = "cheap";
export interface AdapterModelProfileDefinition {
key: AdapterModelProfileKey;
label: string;
description?: string;
adapterConfig: Record<string, unknown>;
source?: "adapter_default" | "discovered";
}
export type AdapterEnvironmentCheckLevel = "info" | "warn" | "error";
export interface AdapterEnvironmentCheck {
@@ -318,6 +329,23 @@ export interface AdapterConfigSchema {
fields: ConfigFieldSchema[];
}
export interface AdapterRuntimeCommandSpec {
/**
* The command Paperclip should execute for this adapter in the current config.
*/
command: string;
/**
* Optional command name/path to probe for availability before launch.
* Defaults to `command` when omitted by the consumer.
*/
detectCommand?: string | null;
/**
* Optional shell snippet that can install or expose the adapter command in a
* fresh remote runtime. It should be idempotent.
*/
installCommand?: string | null;
}
export interface ServerAdapterModule {
type: string;
execute(ctx: AdapterExecutionContext): Promise<AdapterExecutionResult>;
@@ -329,6 +357,8 @@ export interface ServerAdapterModule {
supportsLocalAgentJwt?: boolean;
models?: AdapterModel[];
listModels?: () => Promise<AdapterModel[]>;
modelProfiles?: AdapterModelProfileDefinition[];
listModelProfiles?: () => Promise<AdapterModelProfileDefinition[]>;
/**
* Optional explicit refresh hook for model discovery.
* Use this when the adapter caches discovered models and needs a bypass path
@@ -394,6 +424,11 @@ export interface ServerAdapterModule {
* rather than reading config.paperclipRuntimeSkills.
*/
requiresMaterializedRuntimeSkills?: boolean;
/**
* Optional: describe how this adapter's runtime command should be launched
* and provisioned in fresh remote environments such as sandboxes.
*/
getRuntimeCommandSpec?: (config: Record<string, unknown>) => AdapterRuntimeCommandSpec | null;
}
// ---------------------------------------------------------------------------
@@ -435,6 +470,14 @@ export interface CreateConfigValues {
promptTemplate: string;
model: string;
thinkingEffort: string;
/**
* Optional cheap model profile config for new agents on adapters that
* support model profiles. Persisted under
* `runtimeConfig.modelProfiles.cheap.adapterConfig`, never on the primary
* `adapterConfig`.
*/
cheapModel?: string;
cheapModelEnabled?: boolean;
chrome: boolean;
dangerouslySkipPermissions: boolean;
search: boolean;

View File

@@ -0,0 +1,64 @@
{
"name": "@paperclipai/adapter-acpx-local",
"version": "0.3.1",
"license": "MIT",
"homepage": "https://github.com/paperclipai/paperclip",
"bugs": {
"url": "https://github.com/paperclipai/paperclip/issues"
},
"repository": {
"type": "git",
"url": "https://github.com/paperclipai/paperclip",
"directory": "packages/adapters/acpx-local"
},
"type": "module",
"exports": {
".": "./src/index.ts",
"./server": "./src/server/index.ts",
"./ui": "./src/ui/index.ts",
"./cli": "./src/cli/index.ts"
},
"publishConfig": {
"access": "public",
"exports": {
".": {
"types": "./dist/index.d.ts",
"import": "./dist/index.js"
},
"./server": {
"types": "./dist/server/index.d.ts",
"import": "./dist/server/index.js"
},
"./ui": {
"types": "./dist/ui/index.d.ts",
"import": "./dist/ui/index.js"
},
"./cli": {
"types": "./dist/cli/index.d.ts",
"import": "./dist/cli/index.js"
}
},
"main": "./dist/index.js",
"types": "./dist/index.d.ts"
},
"files": [
"dist",
"skills"
],
"scripts": {
"build": "tsc",
"clean": "rm -rf dist",
"typecheck": "tsc --noEmit"
},
"dependencies": {
"@agentclientprotocol/claude-agent-acp": "^0.31.4",
"@paperclipai/adapter-utils": "workspace:*",
"@zed-industries/codex-acp": "^0.12.0",
"acpx": "^0.6.1",
"picocolors": "^1.1.1"
},
"devDependencies": {
"@types/node": "^24.6.0",
"typescript": "^5.7.3"
}
}

View File

@@ -0,0 +1,121 @@
import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
import { printAcpxStreamEvent } from "./format-event.js";
function emit(payload: Record<string, unknown>): string {
return JSON.stringify(payload);
}
interface CapturedOutput {
log: string[];
stdout: string[];
}
function captureOutput(): { capture: CapturedOutput; restore: () => void } {
const log: string[] = [];
const stdout: string[] = [];
const logSpy = vi.spyOn(console, "log").mockImplementation((value?: unknown) => {
log.push(String(value ?? ""));
});
const stdoutSpy = vi.spyOn(process.stdout, "write").mockImplementation(((chunk: unknown) => {
stdout.push(String(chunk ?? ""));
return true;
}) as typeof process.stdout.write);
return {
capture: { log, stdout },
restore: () => {
logSpy.mockRestore();
stdoutSpy.mockRestore();
},
};
}
function strip(value: string): string {
return value.replace(/\x1b\[[0-9;]*m/g, "");
}
describe("printAcpxStreamEvent", () => {
let captured: CapturedOutput;
let restore: () => void;
beforeEach(() => {
const result = captureOutput();
captured = result.capture;
restore = result.restore;
});
afterEach(() => {
restore();
});
it("renders acpx.session as a labeled session header", () => {
printAcpxStreamEvent(
emit({
type: "acpx.session",
agent: "claude",
acpSessionId: "acp-1",
mode: "persistent",
permissionMode: "approve-all",
}),
false,
);
expect(captured.log.map(strip)).toEqual(["claude session: acp-1 [persistent / approve-all]"]);
});
it("streams output text_delta to stdout for live progress", () => {
printAcpxStreamEvent(
emit({ type: "acpx.text_delta", text: "hello", channel: "output" }),
false,
);
expect(captured.log).toEqual([]);
expect(captured.stdout.map(strip)).toEqual(["hello"]);
});
it("renders thought text_delta on its own line", () => {
printAcpxStreamEvent(
emit({ type: "acpx.text_delta", text: "thinking…", channel: "thought" }),
false,
);
expect(captured.log.map(strip)).toEqual(["thinking…"]);
});
it("renders tool_call with status and id", () => {
printAcpxStreamEvent(
emit({
type: "acpx.tool_call",
name: "read",
toolCallId: "tool-1",
status: "running",
text: "read README.md",
}),
false,
);
expect(captured.log.map(strip)).toEqual([
"tool_call: read [running] (tool-1)",
"read README.md",
]);
});
it("renders status events with optional context window", () => {
printAcpxStreamEvent(
emit({ type: "acpx.status", tag: "context_window", used: 100, size: 200000 }),
false,
);
expect(captured.log.map(strip)).toEqual(["status: context_window (100/200000 ctx)"]);
});
it("renders acpx.result and acpx.error", () => {
printAcpxStreamEvent(emit({ type: "acpx.result", summary: "completed", stopReason: "end_turn" }), false);
printAcpxStreamEvent(emit({ type: "acpx.error", message: "auth required" }), false);
expect(captured.log.map(strip)).toEqual(["result: completed", "error: auth required"]);
});
it("falls back to plain output for non-JSON lines", () => {
printAcpxStreamEvent("not json", false);
expect(captured.log).toEqual(["not json"]);
});
it("still emits unknown / non-JSON lines when debug is enabled", () => {
printAcpxStreamEvent("not json", true);
expect(strip(captured.log[0])).toBe("not json");
});
});

View File

@@ -0,0 +1,121 @@
import pc from "picocolors";
function parseJson(line: string): Record<string, unknown> | null {
try {
const parsed = JSON.parse(line);
if (typeof parsed !== "object" || parsed === null || Array.isArray(parsed)) return null;
return parsed as Record<string, unknown>;
} catch {
return null;
}
}
function asString(value: unknown, fallback = ""): string {
return typeof value === "string" ? value : fallback;
}
function asNumber(value: unknown, fallback = 0): number {
return typeof value === "number" && Number.isFinite(value) ? value : fallback;
}
function stringify(value: unknown): string {
if (typeof value === "string") return value;
if (value === null || value === undefined) return "";
try {
return JSON.stringify(value, null, 2);
} catch {
return String(value);
}
}
function pickToolUseId(parsed: Record<string, unknown>): string {
return (
asString(parsed.toolCallId) ||
asString(parsed.toolUseId) ||
asString(parsed.id)
);
}
function statusLine(parsed: Record<string, unknown>): string {
const text = asString(parsed.text).trim();
const tag = asString(parsed.tag).trim();
const used = asNumber(parsed.used, -1);
const size = asNumber(parsed.size, -1);
const parts: string[] = [];
if (text) parts.push(text);
if (tag && !text) parts.push(tag);
if (used >= 0 && size > 0) parts.push(`(${used}/${size} ctx)`);
return parts.join(" ") || tag || "status";
}
export function printAcpxStreamEvent(raw: string, debug: boolean): void {
const line = raw.trim();
if (!line) return;
const parsed = parseJson(line);
if (!parsed) {
if (debug) console.log(pc.gray(line));
else console.log(line);
return;
}
const type = asString(parsed.type);
if (type === "acpx.session") {
const agent = asString(parsed.agent, "acpx");
const session =
asString(parsed.acpSessionId) ||
asString(parsed.sessionId) ||
asString(parsed.runtimeSessionName);
const mode = asString(parsed.mode);
const permissionMode = asString(parsed.permissionMode);
const tail = [mode, permissionMode].filter(Boolean).join(" / ");
const suffix = tail ? ` [${tail}]` : "";
console.log(pc.blue(`${agent} session${session ? `: ${session}` : ""}${suffix}`));
return;
}
if (type === "acpx.text_delta") {
const text = asString(parsed.text);
if (!text) return;
const channel = asString(parsed.channel) || asString(parsed.stream);
const isThought = channel === "thought" || channel === "thinking";
if (isThought) console.log(pc.gray(text));
else process.stdout.write(pc.green(text));
return;
}
if (type === "acpx.tool_call") {
const name = asString(parsed.name, "acp_tool");
const status = asString(parsed.status);
const id = pickToolUseId(parsed);
const header = status ? `tool_call: ${name} [${status}]` : `tool_call: ${name}`;
const idSuffix = id ? ` (${id})` : "";
const isError = status === "failed" || status === "cancelled";
console.log((isError ? pc.red : pc.yellow)(`${header}${idSuffix}`));
if (parsed.input !== undefined) {
console.log(pc.gray(stringify(parsed.input)));
} else {
const text = asString(parsed.text).trim();
if (text) console.log(pc.gray(text));
}
return;
}
if (type === "acpx.tool_result") {
const isError = parsed.isError === true || parsed.error !== undefined;
console.log((isError ? pc.red : pc.cyan)(`tool_result: ${asString(parsed.name, "acp_tool")}`));
const content = stringify(parsed.content ?? parsed.output ?? parsed.error);
if (content) console.log((isError ? pc.red : pc.gray)(content));
return;
}
if (type === "acpx.status") {
console.log(pc.gray(`status: ${statusLine(parsed)}`));
return;
}
if (type === "acpx.result") {
const summary = asString(parsed.summary, asString(parsed.stopReason, asString(parsed.subtype, "complete")));
console.log(pc.blue(`result: ${summary}`));
return;
}
if (type === "acpx.error") {
console.log(pc.red(`error: ${asString(parsed.message, line)}`));
return;
}
console.log(debug ? pc.gray(line) : line);
}

View File

@@ -0,0 +1 @@
export { printAcpxStreamEvent } from "./format-event.js";

View File

@@ -0,0 +1,47 @@
export const type = "acpx_local";
export const label = "ACPX (local)";
export const DEFAULT_ACPX_LOCAL_AGENT = "claude";
export const DEFAULT_ACPX_LOCAL_MODE = "persistent";
export const DEFAULT_ACPX_LOCAL_PERMISSION_MODE = "approve-all";
export const DEFAULT_ACPX_LOCAL_NON_INTERACTIVE_PERMISSIONS = "deny";
export const DEFAULT_ACPX_LOCAL_TIMEOUT_SEC = 0;
export const acpxAgentOptions = [
{ id: "claude", label: "Claude via ACPX" },
{ id: "codex", label: "Codex via ACPX" },
{ id: "custom", label: "Custom ACP command" },
] as const;
export const agentConfigurationDoc = `# acpx_local agent configuration
Adapter: acpx_local
Use when:
- The agent should run through Agent Client Protocol via ACPX on the Paperclip host or a managed execution environment.
- You want one built-in adapter that can target Claude, Codex, or a custom ACP server command.
- You need Paperclip-managed session identity and live streamed ACP events in later ACPX runtime phases.
Don't use when:
- You need today's stable Claude Code or Codex CLI wrapper behavior. Use claude_local or codex_local until acpx_local runtime execution is enabled.
- The host cannot satisfy ACPX's Node >=22.12.0 prerequisite.
- The agent runtime is not an ACP server and cannot be launched through ACPX.
Core fields:
- agent (string, optional): claude, codex, or custom. Defaults to claude.
- agentCommand (string, optional): custom ACP command when agent=custom, or an override for a built-in ACP agent command.
- mode (string, optional): persistent or oneshot. Defaults to persistent.
- cwd (string, optional): default absolute working directory fallback for the agent process.
- permissionMode (string, optional): defaults to approve-all, meaning ACPX permission requests are auto-approved.
- nonInteractivePermissions (string, optional): fallback behavior when ACPX cannot ask interactively. Supported values are deny and fail.
- stateDir (string, optional): ACPX state directory. Defaults to a Paperclip-managed company/agent scoped location.
- instructionsFilePath (string, optional): absolute path to a markdown instructions file used by Paperclip prompt construction.
- promptTemplate (string, optional): run prompt template.
- bootstrapPromptTemplate (string, optional): first-run bootstrap prompt template.
- timeoutSec (number, optional): run timeout in seconds. Defaults to 0, meaning no adapter timeout.
- env (object, optional): KEY=VALUE environment variables or secret bindings.
Dependency decision:
- acpx_local declares direct dependencies on acpx, @agentclientprotocol/claude-agent-acp, and @zed-industries/codex-acp so the built-in adapter has deterministic package resolution instead of relying on globally installed ACP commands.
- ACPX currently requires Node >=22.12.0. Paperclip keeps the repo-wide Node >=20 engine and surfaces the stricter runtime prerequisite through acpx_local diagnostics.
`;

View File

@@ -0,0 +1,102 @@
import type { AdapterConfigSchema } from "@paperclipai/adapter-utils";
import {
DEFAULT_ACPX_LOCAL_AGENT,
DEFAULT_ACPX_LOCAL_MODE,
DEFAULT_ACPX_LOCAL_NON_INTERACTIVE_PERMISSIONS,
DEFAULT_ACPX_LOCAL_PERMISSION_MODE,
DEFAULT_ACPX_LOCAL_TIMEOUT_SEC,
acpxAgentOptions,
} from "../index.js";
export function getConfigSchema(): AdapterConfigSchema {
return {
fields: [
{
key: "agent",
label: "ACP agent",
type: "select",
default: DEFAULT_ACPX_LOCAL_AGENT,
required: true,
options: acpxAgentOptions.map((agent) => ({ value: agent.id, label: agent.label })),
hint: "Choose the ACP agent launched through ACPX.",
},
{
key: "agentCommand",
label: "Agent command",
type: "text",
hint: "Required for custom agents; optional override for built-in Claude or Codex ACP commands.",
},
{
key: "mode",
label: "Session mode",
type: "select",
default: DEFAULT_ACPX_LOCAL_MODE,
options: [
{ value: "persistent", label: "Persistent" },
{ value: "oneshot", label: "One shot" },
],
},
{
key: "permissionMode",
label: "Permission mode",
type: "select",
default: DEFAULT_ACPX_LOCAL_PERMISSION_MODE,
options: [
{ value: "approve-all", label: "Approve all" },
{ value: "default", label: "Approve reads" },
],
hint: "Defaults to maximum permissions. Approve reads grants read-only requests and asks for approval on writes.",
},
{
key: "nonInteractivePermissions",
label: "Non-interactive permissions",
type: "select",
default: DEFAULT_ACPX_LOCAL_NON_INTERACTIVE_PERMISSIONS,
options: [
{ value: "deny", label: "Deny" },
{ value: "fail", label: "Fail" },
],
},
{
key: "cwd",
label: "Working directory",
type: "text",
hint: "Absolute fallback directory. Paperclip execution workspaces can override this at runtime.",
},
{
key: "stateDir",
label: "State directory",
type: "text",
hint: "Optional ACPX session state directory. Defaults to Paperclip-managed company/agent scoped storage.",
},
{
key: "instructionsFilePath",
label: "Instructions file",
type: "text",
hint: "Optional absolute path to markdown instructions injected into the run prompt.",
},
{
key: "promptTemplate",
label: "Prompt template",
type: "textarea",
},
{
key: "bootstrapPromptTemplate",
label: "Bootstrap prompt template",
type: "textarea",
},
{
key: "timeoutSec",
label: "Timeout seconds",
type: "number",
default: DEFAULT_ACPX_LOCAL_TIMEOUT_SEC,
},
{
key: "env",
label: "Environment JSON",
type: "textarea",
hint: "Optional JSON object of environment values or secret bindings.",
},
],
};
}

View File

@@ -0,0 +1,425 @@
import fs from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import { afterEach, describe, expect, it } from "vitest";
import { createAcpxLocalExecutor } from "./execute.js";
const tempRoots: string[] = [];
async function makeTempRoot() {
const root = await fs.mkdtemp(path.join(os.tmpdir(), "paperclip-acpx-skills-"));
tempRoots.push(root);
return root;
}
afterEach(async () => {
await Promise.all(tempRoots.splice(0).map((root) => fs.rm(root, { recursive: true, force: true })));
});
async function pathExists(candidate: string): Promise<boolean> {
return fs.access(candidate).then(() => true).catch(() => false);
}
async function onlyChildDir(parent: string): Promise<string> {
const entries = await fs.readdir(parent);
expect(entries).toHaveLength(1);
return path.join(parent, entries[0]!);
}
async function createSkill(root: string, name: string, body = `---\nrequired: false\n---\n# ${name}\n`) {
const skillDir = path.join(root, name);
await fs.mkdir(skillDir, { recursive: true });
await fs.writeFile(path.join(skillDir, "SKILL.md"), body, "utf8");
return {
key: `paperclipai/test/${name}`,
runtimeName: name,
source: skillDir,
required: false,
};
}
function buildRuntime() {
return {
ensureSession: async () => ({
backendSessionId: "backend-session",
agentSessionId: "agent-session",
runtimeSessionName: "runtime-session",
}),
startTurn: () => ({
events: (async function* () {
yield { type: "done", stopReason: "end_turn" };
})(),
result: Promise.resolve({ status: "completed", stopReason: "end_turn" }),
cancel: async () => {},
}),
close: async () => {},
};
}
async function runExecutor(
config: Record<string, unknown>,
options: {
context?: Record<string, unknown>;
executionTransport?: Record<string, unknown>;
} = {},
) {
const runtimeOptions: Record<string, unknown>[] = [];
const meta: Record<string, unknown>[] = [];
const logs: Array<{ stream: string; text: string }> = [];
const execute = createAcpxLocalExecutor({
createRuntime: (options) => {
runtimeOptions.push(options as unknown as Record<string, unknown>);
return buildRuntime() as never;
},
});
const result = await execute({
runId: "run-1",
agent: {
id: "agent-1",
companyId: "company-1",
},
runtime: {},
config,
context: options.context ?? {},
executionTransport: options.executionTransport,
onLog: async (stream: "stdout" | "stderr", text: string) => {
logs.push({ stream, text });
},
onMeta: async (payload: unknown) => {
meta.push(payload as Record<string, unknown>);
},
} as never);
expect(result.exitCode).toBe(0);
return { logs, meta, runtimeOptions, result };
}
describe("acpx_local runtime skill isolation", () => {
it.skipIf(process.platform === "win32")("materializes ACPX Claude skills without symlinked descendants", async () => {
const root = await makeTempRoot();
const skillRoot = path.join(root, "skills");
const outsideRoot = path.join(root, "outside");
await fs.mkdir(outsideRoot, { recursive: true });
await fs.writeFile(path.join(outsideRoot, "secret.txt"), "do not expose", "utf8");
const skill = await createSkill(skillRoot, "danger");
await fs.symlink(path.join(outsideRoot, "secret.txt"), path.join(skill.source, "leak.txt"));
await fs.symlink(outsideRoot, path.join(skill.source, "leak-dir"));
const stateDir = path.join(root, "state");
const { meta } = await runExecutor({
agent: "claude",
stateDir,
paperclipRuntimeSkills: [skill],
paperclipSkillSync: { desiredSkills: [skill.key] },
});
const mountedRoot = await onlyChildDir(path.join(stateDir, "runtime-skills", "claude"));
const skillsHome = path.join(mountedRoot, ".claude", "skills");
const materializedSkill = path.join(skillsHome, skill.runtimeName);
expect(await fs.readFile(path.join(materializedSkill, "SKILL.md"), "utf8")).toContain("# danger");
expect(await pathExists(path.join(materializedSkill, "leak.txt"))).toBe(false);
expect(await pathExists(path.join(materializedSkill, "leak-dir"))).toBe(false);
expect(String(meta[0]?.prompt ?? "")).toContain(`Skill root: ${skillsHome}`);
});
it.skipIf(process.platform === "win32")("revokes removed ACPX Codex skills and skips symlinked descendants", async () => {
const root = await makeTempRoot();
const skillRoot = path.join(root, "skills");
const outsideRoot = path.join(root, "outside");
const codexHome = path.join(root, "codex-home");
await fs.mkdir(outsideRoot, { recursive: true });
await fs.writeFile(path.join(outsideRoot, "secret.txt"), "do not expose", "utf8");
const keep = await createSkill(skillRoot, "keep");
const remove = await createSkill(skillRoot, "remove");
await fs.symlink(path.join(outsideRoot, "secret.txt"), path.join(keep.source, "leak.txt"));
await fs.symlink(outsideRoot, path.join(keep.source, "leak-dir"));
const baseConfig = {
agent: "codex",
stateDir: path.join(root, "state"),
env: { CODEX_HOME: codexHome },
paperclipRuntimeSkills: [keep, remove],
};
await runExecutor({
...baseConfig,
paperclipSkillSync: { desiredSkills: [keep.key, remove.key] },
});
expect(await pathExists(path.join(codexHome, "skills", remove.runtimeName, "SKILL.md"))).toBe(true);
await runExecutor({
...baseConfig,
paperclipSkillSync: { desiredSkills: [keep.key] },
});
expect(await pathExists(path.join(codexHome, "skills", keep.runtimeName, "SKILL.md"))).toBe(true);
expect(await pathExists(path.join(codexHome, "skills", keep.runtimeName, "leak.txt"))).toBe(false);
expect(await pathExists(path.join(codexHome, "skills", keep.runtimeName, "leak-dir"))).toBe(false);
expect(await pathExists(path.join(codexHome, "skills", remove.runtimeName))).toBe(false);
});
it.skipIf(process.platform === "win32")("removes legacy ACPX Codex skill symlinks when a skill is no longer desired", async () => {
const root = await makeTempRoot();
const skillRoot = path.join(root, "skills");
const codexHome = path.join(root, "codex-home");
const legacy = await createSkill(skillRoot, "legacy");
const skillsHome = path.join(codexHome, "skills");
await fs.mkdir(skillsHome, { recursive: true });
await fs.symlink(legacy.source, path.join(skillsHome, legacy.runtimeName));
await runExecutor({
agent: "codex",
stateDir: path.join(root, "state"),
env: { CODEX_HOME: codexHome },
paperclipRuntimeSkills: [legacy],
paperclipSkillSync: { desiredSkills: [] },
});
expect(await pathExists(path.join(skillsHome, legacy.runtimeName))).toBe(false);
});
it.skipIf(process.platform === "win32")("replaces stale managed Codex auth files with source symlinks", async () => {
const root = await makeTempRoot();
const sourceCodexHome = path.join(root, "source-codex-home");
const paperclipHome = path.join(root, "paperclip-home");
const paperclipInstanceId = "test-instance";
const managedCodexHome = path.join(
paperclipHome,
"instances",
paperclipInstanceId,
"companies",
"company-1",
"codex-home",
);
await fs.mkdir(sourceCodexHome, { recursive: true });
await fs.mkdir(managedCodexHome, { recursive: true });
const sourceAuth = path.join(sourceCodexHome, "auth.json");
const managedAuth = path.join(managedCodexHome, "auth.json");
await fs.writeFile(sourceAuth, "{\"source\":true}", "utf8");
await fs.writeFile(managedAuth, "{\"stale\":true}", "utf8");
const previousCodexHome = process.env.CODEX_HOME;
const previousPaperclipHome = process.env.PAPERCLIP_HOME;
const previousPaperclipInstanceId = process.env.PAPERCLIP_INSTANCE_ID;
try {
process.env.CODEX_HOME = sourceCodexHome;
process.env.PAPERCLIP_HOME = paperclipHome;
process.env.PAPERCLIP_INSTANCE_ID = paperclipInstanceId;
await runExecutor({
agent: "codex",
stateDir: path.join(root, "state"),
paperclipRuntimeSkills: [],
paperclipSkillSync: { desiredSkills: [] },
});
} finally {
if (previousCodexHome === undefined) delete process.env.CODEX_HOME;
else process.env.CODEX_HOME = previousCodexHome;
if (previousPaperclipHome === undefined) delete process.env.PAPERCLIP_HOME;
else process.env.PAPERCLIP_HOME = previousPaperclipHome;
if (previousPaperclipInstanceId === undefined) delete process.env.PAPERCLIP_INSTANCE_ID;
else process.env.PAPERCLIP_INSTANCE_ID = previousPaperclipInstanceId;
}
const authStat = await fs.lstat(managedAuth);
expect(authStat.isSymbolicLink()).toBe(true);
expect(path.resolve(path.dirname(managedAuth), await fs.readlink(managedAuth))).toBe(sourceAuth);
});
it("keeps fresh credential wrapper scripts across ACPX agent changes", async () => {
const root = await makeTempRoot();
const stateDir = path.join(root, "state");
const baseConfig = {
agentCommand: "node ./fake-acp.js",
stateDir,
};
await runExecutor({
...baseConfig,
agent: "custom-a",
env: { PAPERCLIP_API_KEY: "old-key" },
});
await runExecutor({
...baseConfig,
agent: "custom-b",
env: { PAPERCLIP_API_KEY: "new-key" },
});
const wrappers = await fs.readdir(path.join(stateDir, "wrappers"));
expect(wrappers.filter((name) => name.endsWith(".sh"))).toHaveLength(2);
expect(wrappers.filter((name) => name.endsWith(".env"))).toHaveLength(2);
expect(wrappers.some((name) => name.startsWith("custom-a-"))).toBe(true);
expect(wrappers.some((name) => name.startsWith("custom-b-"))).toBe(true);
const wrapperPath = path.join(stateDir, "wrappers", wrappers.find((name) => name.startsWith("custom-b-") && name.endsWith(".sh"))!);
const envPath = path.join(stateDir, "wrappers", wrappers.find((name) => name.startsWith("custom-b-") && name.endsWith(".env"))!);
const wrapper = await fs.readFile(wrapperPath, "utf8");
const env = await fs.readFile(envPath, "utf8");
expect((await fs.stat(envPath)).mode & 0o777).toBe(0o600);
expect((await fs.stat(wrapperPath)).mode & 0o777).toBe(0o700);
expect(wrapper).toContain("node ./fake-acp.js");
expect(wrapper).not.toContain("PAPERCLIP_API_KEY");
expect(wrapper).not.toContain("new-key");
expect(wrapper).not.toContain("old-key");
expect(env).toContain("PAPERCLIP_API_KEY='new-key'");
expect(env).not.toContain("old-key");
});
it("shapes ACPX wrapper workspace env for remote execution identities", async () => {
const root = await makeTempRoot();
const stateDir = path.join(root, "state");
const workspaceDir = path.join(root, "workspace");
await fs.mkdir(workspaceDir, { recursive: true });
await runExecutor(
{
agentCommand: "node ./fake-acp.js",
stateDir,
},
{
context: {
paperclipWorkspace: {
cwd: workspaceDir,
source: "project_primary",
strategy: "git_worktree",
workspaceId: "workspace-1",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
branchName: "feature/remote-acpx",
worktreePath: workspaceDir,
},
},
executionTransport: {
remoteExecution: {
host: "127.0.0.1",
port: 2222,
username: "fixture",
remoteWorkspacePath: "/remote/workspace",
remoteCwd: "/remote/workspace",
privateKey: "PRIVATE KEY",
knownHosts: "[127.0.0.1]:2222 ssh-ed25519 AAAA",
strictHostKeyChecking: true,
},
},
},
);
const wrappers = await fs.readdir(path.join(stateDir, "wrappers"));
const envPath = path.join(
stateDir,
"wrappers",
wrappers.find((name) => name.endsWith(".env"))!,
);
const env = await fs.readFile(envPath, "utf8");
expect(env).toContain("PAPERCLIP_WORKSPACE_CWD='/remote/workspace'");
expect(env).not.toContain("PAPERCLIP_WORKSPACE_WORKTREE_PATH=");
});
it("cleans aged credential wrapper scripts across ACPX agent changes", async () => {
const root = await makeTempRoot();
const stateDir = path.join(root, "state");
const wrappersDir = path.join(stateDir, "wrappers");
const baseConfig = {
agentCommand: "node ./fake-acp.js",
stateDir,
};
await runExecutor({
...baseConfig,
agent: "custom-a",
env: { PAPERCLIP_API_KEY: "old-key" },
});
const oldDate = new Date(Date.now() - 16 * 60 * 1000);
await Promise.all(
(await fs.readdir(wrappersDir))
.filter((name) => name.startsWith("custom-a-"))
.map((name) => fs.utimes(path.join(wrappersDir, name), oldDate, oldDate)),
);
await runExecutor({
...baseConfig,
agent: "custom-b",
env: { PAPERCLIP_API_KEY: "new-key" },
});
const wrappers = await fs.readdir(wrappersDir);
expect(wrappers.filter((name) => name.endsWith(".sh"))).toHaveLength(1);
expect(wrappers.filter((name) => name.endsWith(".env"))).toHaveLength(1);
expect(wrappers.some((name) => name.startsWith("custom-a-"))).toBe(false);
expect(wrappers.some((name) => name.startsWith("custom-b-"))).toBe(true);
});
it("keeps distinct wrapper env files for concurrent runs with different credentials", async () => {
const root = await makeTempRoot();
const stateDir = path.join(root, "state");
const baseConfig = {
agent: "custom-a",
agentCommand: "node ./fake-acp.js",
stateDir,
};
await runExecutor({
...baseConfig,
env: { PAPERCLIP_API_KEY: "first-key" },
});
await runExecutor({
...baseConfig,
env: { PAPERCLIP_API_KEY: "second-key" },
});
const envFileNames = (await fs.readdir(path.join(stateDir, "wrappers"))).filter((name) => name.endsWith(".env"));
expect(envFileNames).toHaveLength(2);
const envFiles = await Promise.all(
envFileNames.map(async (name) => fs.readFile(path.join(stateDir, "wrappers", name), "utf8")),
);
expect(envFiles.filter((contents) => contents.includes("PAPERCLIP_API_KEY='first-key'"))).toHaveLength(1);
expect(envFiles.filter((contents) => contents.includes("PAPERCLIP_API_KEY='second-key'"))).toHaveLength(1);
});
it("passes Paperclip env through the ACP agent wrapper instead of process.env", async () => {
let observedApiKeyDuringStream: string | undefined;
const execute = createAcpxLocalExecutor({
createRuntime: () => ({
ensureSession: async () => ({
backendSessionId: "backend-session",
agentSessionId: "agent-session",
runtimeSessionName: "runtime-session",
}),
startTurn: () => ({
events: (async function* () {
await Promise.resolve();
observedApiKeyDuringStream = process.env.PAPERCLIP_API_KEY;
yield { type: "done", stopReason: "end_turn" };
})(),
result: Promise.resolve({ status: "completed", stopReason: "end_turn" }),
cancel: async () => {},
}),
close: async () => {},
}) as never,
});
const previousApiKey = process.env.PAPERCLIP_API_KEY;
try {
delete process.env.PAPERCLIP_API_KEY;
const result = await execute({
runId: "run-1",
agent: {
id: "agent-1",
companyId: "company-1",
},
runtime: {},
config: { agent: "custom", agentCommand: "node ./fake-acp.js" },
context: {},
authToken: "runtime-key",
onLog: async () => {},
onMeta: async () => {},
} as never);
expect(result.exitCode).toBe(0);
expect(observedApiKeyDuringStream).toBeUndefined();
} finally {
if (previousApiKey === undefined) delete process.env.PAPERCLIP_API_KEY;
else process.env.PAPERCLIP_API_KEY = previousApiKey;
}
});
});

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,5 @@
export { execute, createAcpxLocalExecutor } from "./execute.js";
export { testEnvironment } from "./test.js";
export { getConfigSchema } from "./config-schema.js";
export { sessionCodec } from "./session-codec.js";
export { listAcpxSkills, syncAcpxSkills } from "./skills.js";

View File

@@ -0,0 +1,50 @@
import type { AdapterSessionCodec } from "@paperclipai/adapter-utils";
function readString(value: unknown): string | null {
return typeof value === "string" && value.trim().length > 0 ? value.trim() : null;
}
function readRecord(value: unknown): Record<string, unknown> | null {
return typeof value === "object" && value !== null && !Array.isArray(value) ? { ...(value as Record<string, unknown>) } : null;
}
export const sessionCodec: AdapterSessionCodec = {
deserialize(raw: unknown) {
if (typeof raw !== "object" || raw === null || Array.isArray(raw)) return null;
const record = raw as Record<string, unknown>;
const runtimeSessionName = readString(record.runtimeSessionName);
const acpSessionId = readString(record.acpSessionId);
const agentSessionId = readString(record.agentSessionId);
const remoteExecution = readRecord(record.remoteExecution);
if (!runtimeSessionName && !acpSessionId && !agentSessionId) return null;
return {
...(runtimeSessionName ? { runtimeSessionName } : {}),
...(readString(record.sessionKey) ? { sessionKey: readString(record.sessionKey) } : {}),
...(readString(record.acpxRecordId) ? { acpxRecordId: readString(record.acpxRecordId) } : {}),
...(acpSessionId ? { acpSessionId } : {}),
...(agentSessionId ? { agentSessionId } : {}),
...(readString(record.agent) ? { agent: readString(record.agent) } : {}),
...(readString(record.cwd) ? { cwd: readString(record.cwd) } : {}),
...(readString(record.mode) ? { mode: readString(record.mode) } : {}),
...(readString(record.stateDir) ? { stateDir: readString(record.stateDir) } : {}),
...(readString(record.configFingerprint) ? { configFingerprint: readString(record.configFingerprint) } : {}),
...(readString(record.workspaceId) ? { workspaceId: readString(record.workspaceId) } : {}),
...(readString(record.repoUrl) ? { repoUrl: readString(record.repoUrl) } : {}),
...(readString(record.repoRef) ? { repoRef: readString(record.repoRef) } : {}),
...(remoteExecution ? { remoteExecution } : {}),
};
},
serialize(params: Record<string, unknown> | null) {
if (!params) return null;
return this.deserialize(params);
},
getDisplayId(params: Record<string, unknown> | null) {
if (!params) return null;
return (
readString(params.runtimeSessionName) ??
readString(params.acpSessionId) ??
readString(params.agentSessionId)
);
},
};

View File

@@ -0,0 +1,106 @@
import path from "node:path";
import { fileURLToPath } from "node:url";
import type {
AdapterSkillContext,
AdapterSkillEntry,
AdapterSkillSnapshot,
} from "@paperclipai/adapter-utils";
import {
readPaperclipRuntimeSkillEntries,
resolvePaperclipDesiredSkillNames,
} from "@paperclipai/adapter-utils/server-utils";
const __moduleDir = path.dirname(fileURLToPath(import.meta.url));
type AcpxSkillAgent = "claude" | "codex" | "custom";
function normalizeAcpxSkillAgent(config: Record<string, unknown>): AcpxSkillAgent {
const configured = typeof config.agent === "string" ? config.agent.trim() : "";
if (configured === "codex" || configured === "custom") return configured;
if (configured === "claude" || configured === "") return "claude";
return "claude";
}
function configuredDetail(agent: AcpxSkillAgent): string {
if (agent === "codex") {
return "Will be linked into the effective CODEX_HOME/skills/ directory for the next ACPX Codex session.";
}
return "Will be mounted into the next ACPX Claude session.";
}
function unsupportedDetail(): string {
return "Desired state is stored in Paperclip only; custom ACP commands need an explicit skill integration contract before runtime sync is available.";
}
async function buildAcpxSkillSnapshot(config: Record<string, unknown>): Promise<AdapterSkillSnapshot> {
const acpxAgent = normalizeAcpxSkillAgent(config);
const availableEntries = await readPaperclipRuntimeSkillEntries(config, __moduleDir);
const availableByKey = new Map(availableEntries.map((entry) => [entry.key, entry]));
const desiredSkills = resolvePaperclipDesiredSkillNames(config, availableEntries);
const desiredSet = new Set(desiredSkills);
const supported = acpxAgent !== "custom";
const warnings: string[] = supported
? []
: [
"Custom ACP commands do not expose a Paperclip skill integration contract yet; selected skills are tracked only.",
];
const entries: AdapterSkillEntry[] = availableEntries.map((entry) => {
const desired = desiredSet.has(entry.key);
return {
key: entry.key,
runtimeName: entry.runtimeName,
desired,
managed: true,
state: desired ? "configured" : "available",
origin: entry.required ? "paperclip_required" : "company_managed",
originLabel: entry.required ? "Required by Paperclip" : "Managed by Paperclip",
readOnly: false,
sourcePath: entry.source,
targetPath: null,
detail: desired ? (supported ? configuredDetail(acpxAgent) : unsupportedDetail()) : null,
required: Boolean(entry.required),
requiredReason: entry.requiredReason ?? null,
};
});
for (const desiredSkill of desiredSkills) {
if (availableByKey.has(desiredSkill)) continue;
warnings.push(`Desired skill "${desiredSkill}" is not available from the Paperclip skills directory.`);
entries.push({
key: desiredSkill,
runtimeName: null,
desired: true,
managed: true,
state: "missing",
origin: "external_unknown",
originLabel: "External or unavailable",
readOnly: false,
sourcePath: null,
targetPath: null,
detail: "Paperclip cannot find this skill in the local runtime skills directory.",
});
}
entries.sort((left, right) => left.key.localeCompare(right.key));
return {
adapterType: "acpx_local",
supported,
mode: supported ? "ephemeral" : "unsupported",
desiredSkills,
entries,
warnings,
};
}
export async function listAcpxSkills(ctx: AdapterSkillContext): Promise<AdapterSkillSnapshot> {
return buildAcpxSkillSnapshot(ctx.config);
}
export async function syncAcpxSkills(
ctx: AdapterSkillContext,
_desiredSkills: string[],
): Promise<AdapterSkillSnapshot> {
return buildAcpxSkillSnapshot(ctx.config);
}

View File

@@ -0,0 +1,49 @@
import { afterEach, describe, expect, it } from "vitest";
import { testEnvironment } from "./test.js";
const originalNodeVersion = process.version;
function setNodeVersion(version: string): void {
Object.defineProperty(process, "version", {
configurable: true,
enumerable: true,
value: version,
});
}
afterEach(() => {
setNodeVersion(originalNodeVersion);
});
describe("acpx_local environment diagnostics", () => {
it("does not force healthy default Claude diagnostics to warn", async () => {
setNodeVersion("v22.12.0");
const result = await testEnvironment({
adapterType: "acpx_local",
companyId: "test-company",
config: { agent: "claude" },
});
expect(result.status).toBe("pass");
expect(result.checks).toContainEqual(
expect.objectContaining({
code: "acpx_agent_selected",
level: "info",
message: "ACP agent selected: claude",
}),
);
expect(result.checks).toContainEqual(
expect.objectContaining({
code: "acpx_runtime_scaffold",
level: "info",
}),
);
expect(result.checks).not.toContainEqual(
expect.objectContaining({
code: "acpx_runtime_scaffold",
level: "warn",
}),
);
});
});

View File

@@ -0,0 +1,295 @@
import { createRequire } from "node:module";
import fs from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import type {
AdapterEnvironmentCheck,
AdapterEnvironmentTestContext,
AdapterEnvironmentTestResult,
} from "@paperclipai/adapter-utils";
import {
asString,
parseObject,
} from "@paperclipai/adapter-utils/server-utils";
const require = createRequire(import.meta.url);
const MIN_NODE_MAJOR = 22;
const MIN_NODE_MINOR = 12;
const MIN_NODE_PATCH = 0;
function summarizeStatus(checks: AdapterEnvironmentCheck[]): AdapterEnvironmentTestResult["status"] {
if (checks.some((check) => check.level === "error")) return "fail";
if (checks.some((check) => check.level === "warn")) return "warn";
return "pass";
}
function nodeVersionMeetsMinimum(version: string): boolean {
const [major = 0, minor = 0, patch = 0] = version
.replace(/^v/, "")
.split(".")
.map((part) => Number.parseInt(part, 10));
if (major > MIN_NODE_MAJOR) return true;
if (major < MIN_NODE_MAJOR) return false;
if (minor > MIN_NODE_MINOR) return true;
if (minor < MIN_NODE_MINOR) return false;
return patch >= MIN_NODE_PATCH;
}
function isNonEmpty(value: unknown): value is string {
return typeof value === "string" && value.trim().length > 0;
}
function getStringEnv(configEnv: Record<string, string>, key: string): string | undefined {
const configured = configEnv[key];
if (typeof configured === "string") return configured;
return process.env[key];
}
function credentialSource(configEnv: Record<string, string>, key: string): string {
return typeof configEnv[key] === "string" ? "adapter config env" : "server environment";
}
async function readJsonObject(filePath: string): Promise<Record<string, unknown> | null> {
try {
const parsed = JSON.parse(await fs.readFile(filePath, "utf8")) as unknown;
return typeof parsed === "object" && parsed !== null && !Array.isArray(parsed)
? parsed as Record<string, unknown>
: null;
} catch {
return null;
}
}
function readNestedString(record: Record<string, unknown>, pathSegments: string[]): string | null {
let current: unknown = record;
for (const segment of pathSegments) {
if (typeof current !== "object" || current === null || Array.isArray(current)) return null;
current = (current as Record<string, unknown>)[segment];
}
return isNonEmpty(current) ? current.trim() : null;
}
async function hasClaudeSubscriptionCredentials(configDir: string): Promise<boolean> {
for (const filename of [".credentials.json", "credentials.json"]) {
const credentials = await readJsonObject(path.join(configDir, filename));
if (!credentials) continue;
if (readNestedString(credentials, ["claudeAiOauth", "accessToken"])) return true;
}
return false;
}
async function hasCodexNativeCredentials(codexHome: string): Promise<boolean> {
const auth = await readJsonObject(path.join(codexHome, "auth.json"));
if (!auth) return false;
return Boolean(
readNestedString(auth, ["accessToken"]) ||
readNestedString(auth, ["tokens", "access_token"]) ||
readNestedString(auth, ["OPENAI_API_KEY"]),
);
}
async function buildCredentialHintChecks(
agent: string,
configEnv: Record<string, string>,
): Promise<AdapterEnvironmentCheck[]> {
if (agent === "claude") {
const bedrockFlag = getStringEnv(configEnv, "CLAUDE_CODE_USE_BEDROCK");
const bedrockBaseUrl = getStringEnv(configEnv, "ANTHROPIC_BEDROCK_BASE_URL");
const hasBedrock =
bedrockFlag === "1" ||
/^true$/i.test(bedrockFlag ?? "") ||
isNonEmpty(bedrockBaseUrl);
const bedrockSourceKey = isNonEmpty(bedrockFlag)
? "CLAUDE_CODE_USE_BEDROCK"
: "ANTHROPIC_BEDROCK_BASE_URL";
const anthropicApiKey = getStringEnv(configEnv, "ANTHROPIC_API_KEY");
const claudeConfigDir = isNonEmpty(getStringEnv(configEnv, "CLAUDE_CONFIG_DIR"))
? path.resolve(getStringEnv(configEnv, "CLAUDE_CONFIG_DIR") as string)
: path.join(os.homedir(), ".claude");
if (hasBedrock) {
return [{
code: "acpx_claude_bedrock_auth_detected",
level: "info",
message: "Claude credential hint: Bedrock auth indicators are configured.",
detail: `Detected in ${credentialSource(configEnv, bedrockSourceKey)}.`,
hint: "Ensure AWS credentials and AWS_REGION are available to the ACPX-launched Claude agent.",
}];
}
if (isNonEmpty(anthropicApiKey)) {
return [{
code: "acpx_claude_anthropic_api_key_detected",
level: "info",
message: "Claude credential hint: ANTHROPIC_API_KEY is set.",
detail: `Detected in ${credentialSource(configEnv, "ANTHROPIC_API_KEY")}.`,
}];
}
if (await hasClaudeSubscriptionCredentials(claudeConfigDir)) {
return [{
code: "acpx_claude_subscription_auth_detected",
level: "info",
message: "Claude credential hint: local Claude subscription credentials were found.",
detail: `Credentials found in ${claudeConfigDir}.`,
}];
}
return [{
code: "acpx_claude_credentials_missing",
level: "info",
message: "Claude credential hint: no Claude API, Bedrock, or local subscription credentials were detected.",
hint: "Set ANTHROPIC_API_KEY, configure Bedrock, or run `claude login` before starting an ACPX Claude agent.",
}];
}
if (agent === "codex") {
const openAiApiKey = getStringEnv(configEnv, "OPENAI_API_KEY");
const codexHome = isNonEmpty(getStringEnv(configEnv, "CODEX_HOME"))
? path.resolve(getStringEnv(configEnv, "CODEX_HOME") as string)
: path.join(os.homedir(), ".codex");
if (isNonEmpty(openAiApiKey)) {
return [{
code: "acpx_codex_openai_api_key_detected",
level: "info",
message: "Codex credential hint: OPENAI_API_KEY is set.",
detail: `Detected in ${credentialSource(configEnv, "OPENAI_API_KEY")}.`,
}];
}
if (await hasCodexNativeCredentials(codexHome)) {
return [{
code: "acpx_codex_native_auth_detected",
level: "info",
message: "Codex credential hint: local Codex auth configuration was found.",
detail: `Credentials found in ${path.join(codexHome, "auth.json")}.`,
}];
}
return [{
code: "acpx_codex_credentials_missing",
level: "info",
message: "Codex credential hint: no OpenAI API key or local Codex auth configuration was detected.",
hint: "Set OPENAI_API_KEY or run `codex login` before starting an ACPX Codex agent.",
}];
}
return [];
}
function resolvePackage(name: string): AdapterEnvironmentCheck {
try {
const resolved = require.resolve(`${name}/package.json`);
return {
code: `acpx_package_${name.replace(/[^a-z0-9]+/gi, "_").toLowerCase()}_present`,
level: "info",
message: `${name} is resolvable.`,
detail: resolved,
};
} catch {
return {
code: `acpx_package_${name.replace(/[^a-z0-9]+/gi, "_").toLowerCase()}_missing`,
level: "error",
message: `${name} is not resolvable from the acpx_local adapter package.`,
hint: "Run pnpm install so the ACPX adapter dependencies are installed.",
};
}
}
async function checkDirectory(pathValue: string, code: string, label: string): Promise<AdapterEnvironmentCheck | null> {
const dir = pathValue.trim();
if (!dir) return null;
try {
await fs.mkdir(dir, { recursive: true });
await fs.access(dir);
return {
code,
level: "info",
message: `${label} is writable: ${dir}`,
};
} catch (err) {
return {
code: `${code}_invalid`,
level: "error",
message: err instanceof Error ? err.message : `${label} is not writable.`,
detail: dir,
};
}
}
export async function testEnvironment(
ctx: AdapterEnvironmentTestContext,
): Promise<AdapterEnvironmentTestResult> {
const config = parseObject(ctx.config);
const envConfig = parseObject(config.env);
const configEnv: Record<string, string> = {};
for (const [key, value] of Object.entries(envConfig)) {
if (typeof value === "string") configEnv[key] = value;
}
const checks: AdapterEnvironmentCheck[] = [];
const nodeVersion = process.version;
checks.push({
code: nodeVersionMeetsMinimum(nodeVersion) ? "acpx_node_supported" : "acpx_node_unsupported",
level: nodeVersionMeetsMinimum(nodeVersion) ? "info" : "error",
message: nodeVersionMeetsMinimum(nodeVersion)
? `Node ${nodeVersion} satisfies ACPX's >=22.12.0 requirement.`
: `Node ${nodeVersion} does not satisfy ACPX's >=22.12.0 requirement.`,
hint: nodeVersionMeetsMinimum(nodeVersion)
? undefined
: "Run acpx_local agents with Node >=22.12.0 or use claude_local/codex_local on Node 20.",
});
checks.push(resolvePackage("acpx"));
checks.push(resolvePackage("@agentclientprotocol/claude-agent-acp"));
checks.push(resolvePackage("@zed-industries/codex-acp"));
const agent = asString(config.agent, "claude");
if (!["claude", "codex", "custom"].includes(agent)) {
checks.push({
code: "acpx_agent_invalid",
level: "error",
message: `Unsupported ACP agent: ${agent}`,
hint: "Use agent=claude, agent=codex, or agent=custom.",
});
} else {
checks.push({
code: "acpx_agent_selected",
level: "info",
message: `ACP agent selected: ${agent}`,
});
checks.push(...await buildCredentialHintChecks(agent, configEnv));
}
if (agent === "custom" && !asString(config.agentCommand, "")) {
checks.push({
code: "acpx_custom_command_missing",
level: "error",
message: "agentCommand is required when agent=custom.",
});
}
const stateDirCheck = await checkDirectory(asString(config.stateDir, ""), "acpx_state_dir_writable", "ACPX state directory");
if (stateDirCheck) checks.push(stateDirCheck);
const permissionMode = asString(config.permissionMode, "approve-all");
checks.push({
code: "acpx_permission_mode",
level: "info",
message: `Effective permission mode: ${permissionMode || "approve-all"}`,
});
checks.push({
code: "acpx_runtime_scaffold",
level: "info",
message: "acpx_local runtime execution is available through the bundled ACPX runtime.",
});
return {
adapterType: ctx.adapterType,
status: summarizeStatus(checks),
checks,
testedAt: new Date().toISOString(),
};
}

View File

@@ -0,0 +1,139 @@
import type { CreateConfigValues } from "@paperclipai/adapter-utils";
import {
DEFAULT_ACPX_LOCAL_AGENT,
DEFAULT_ACPX_LOCAL_MODE,
DEFAULT_ACPX_LOCAL_NON_INTERACTIVE_PERMISSIONS,
DEFAULT_ACPX_LOCAL_PERMISSION_MODE,
DEFAULT_ACPX_LOCAL_TIMEOUT_SEC,
} from "../index.js";
function parseCommaArgs(value: string): string[] {
return value
.split(",")
.map((item) => item.trim())
.filter(Boolean);
}
function parseEnvVars(text: string): Record<string, string> {
const env: Record<string, string> = {};
for (const line of text.split(/\r?\n/)) {
const trimmed = line.trim();
if (!trimmed || trimmed.startsWith("#")) continue;
const eq = trimmed.indexOf("=");
if (eq <= 0) continue;
const key = trimmed.slice(0, eq).trim();
const value = trimmed.slice(eq + 1);
if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(key)) continue;
env[key] = value;
}
return env;
}
function parseEnvBindings(bindings: unknown): Record<string, unknown> {
if (typeof bindings !== "object" || bindings === null || Array.isArray(bindings)) return {};
const env: Record<string, unknown> = {};
for (const [key, raw] of Object.entries(bindings)) {
if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(key)) continue;
if (typeof raw === "string") {
env[key] = { type: "plain", value: raw };
continue;
}
if (typeof raw !== "object" || raw === null || Array.isArray(raw)) continue;
const rec = raw as Record<string, unknown>;
if (rec.type === "plain" && typeof rec.value === "string") {
env[key] = { type: "plain", value: rec.value };
continue;
}
if (rec.type === "secret_ref" && typeof rec.secretId === "string") {
env[key] = {
type: "secret_ref",
secretId: rec.secretId,
...(typeof rec.version === "number" || rec.version === "latest"
? { version: rec.version }
: {}),
};
}
}
return env;
}
function parseJsonObject(text: string): Record<string, unknown> | null {
const trimmed = text.trim();
if (!trimmed) return null;
try {
const parsed = JSON.parse(trimmed);
if (typeof parsed !== "object" || parsed === null || Array.isArray(parsed)) return null;
return parsed as Record<string, unknown>;
} catch {
return null;
}
}
function readNumber(value: unknown, fallback: number): number {
if (typeof value === "number" && Number.isFinite(value)) return value;
if (typeof value === "string" && value.trim()) {
const parsed = Number(value);
if (Number.isFinite(parsed)) return parsed;
}
return fallback;
}
export function buildAcpxLocalConfig(v: CreateConfigValues): Record<string, unknown> {
const schemaValues = v.adapterSchemaValues ?? {};
const ac: Record<string, unknown> = {
agent: schemaValues.agent || DEFAULT_ACPX_LOCAL_AGENT,
mode: schemaValues.mode || DEFAULT_ACPX_LOCAL_MODE,
permissionMode: schemaValues.permissionMode || DEFAULT_ACPX_LOCAL_PERMISSION_MODE,
nonInteractivePermissions:
schemaValues.nonInteractivePermissions || DEFAULT_ACPX_LOCAL_NON_INTERACTIVE_PERMISSIONS,
timeoutSec: readNumber(schemaValues.timeoutSec, DEFAULT_ACPX_LOCAL_TIMEOUT_SEC),
};
for (const key of [
"agentCommand",
"cwd",
"stateDir",
"instructionsFilePath",
"promptTemplate",
"bootstrapPromptTemplate",
]) {
const value = schemaValues[key];
if (typeof value === "string" && value.trim()) ac[key] = value.trim();
}
if (!ac.cwd && v.cwd) ac.cwd = v.cwd;
if (!ac.instructionsFilePath && v.instructionsFilePath) ac.instructionsFilePath = v.instructionsFilePath;
if (!ac.promptTemplate && v.promptTemplate) ac.promptTemplate = v.promptTemplate;
if (!ac.bootstrapPromptTemplate && v.bootstrapPrompt) ac.bootstrapPromptTemplate = v.bootstrapPrompt;
const env = parseEnvBindings(v.envBindings);
const legacy = parseEnvVars(v.envVars);
for (const [key, value] of Object.entries(legacy)) {
if (!Object.prototype.hasOwnProperty.call(env, key)) {
env[key] = { type: "plain", value };
}
}
if (typeof schemaValues.env === "string") {
const schemaEnv = parseJsonObject(schemaValues.env);
if (schemaEnv) Object.assign(env, schemaEnv);
} else if (typeof schemaValues.env === "object" && schemaValues.env !== null && !Array.isArray(schemaValues.env)) {
Object.assign(env, schemaValues.env as Record<string, unknown>);
}
if (Object.keys(env).length > 0) ac.env = env;
if (v.workspaceStrategyType === "git_worktree") {
ac.workspaceStrategy = {
type: "git_worktree",
...(v.workspaceBaseRef ? { baseRef: v.workspaceBaseRef } : {}),
...(v.workspaceBranchTemplate ? { branchTemplate: v.workspaceBranchTemplate } : {}),
...(v.worktreeParentDir ? { worktreeParentDir: v.worktreeParentDir } : {}),
};
}
const runtimeServices = parseJsonObject(v.runtimeServicesJson ?? "");
if (runtimeServices && Array.isArray(runtimeServices.services)) {
ac.workspaceRuntime = runtimeServices;
}
if (v.command) ac.command = v.command;
if (v.extraArgs) ac.extraArgs = parseCommaArgs(v.extraArgs);
return ac;
}

View File

@@ -0,0 +1,2 @@
export { parseAcpxStdoutLine } from "./parse-stdout.js";
export { buildAcpxLocalConfig } from "./build-config.js";

View File

@@ -0,0 +1,160 @@
import { describe, expect, it } from "vitest";
import { parseAcpxStdoutLine } from "./parse-stdout.js";
const TS = "2026-04-30T00:00:00.000Z";
function emit(payload: Record<string, unknown>): string {
return JSON.stringify(payload);
}
describe("parseAcpxStdoutLine", () => {
it("renders an init entry from acpx.session", () => {
const entries = parseAcpxStdoutLine(
emit({
type: "acpx.session",
agent: "claude",
acpSessionId: "acp-1",
runtimeSessionName: "runtime-1",
mode: "persistent",
permissionMode: "approve-all",
}),
TS,
);
expect(entries).toEqual([
{
kind: "init",
ts: TS,
model: "claude (persistent / approve-all)",
sessionId: "acp-1",
},
]);
});
it("routes output text_delta to the assistant transcript", () => {
const entries = parseAcpxStdoutLine(
emit({ type: "acpx.text_delta", text: "hello", channel: "output", tag: "agent_message_chunk" }),
TS,
);
expect(entries).toEqual([
{ kind: "assistant", ts: TS, text: "hello", delta: true },
]);
});
it("routes thought text_delta to the thinking transcript", () => {
const entries = parseAcpxStdoutLine(
emit({ type: "acpx.text_delta", text: "thinking…", channel: "thought" }),
TS,
);
expect(entries).toEqual([
{ kind: "thinking", ts: TS, text: "thinking…", delta: true },
]);
});
it("falls back to stream when channel is missing", () => {
const entries = parseAcpxStdoutLine(
emit({ type: "acpx.text_delta", text: "thinking…", stream: "thought" }),
TS,
);
expect(entries[0]).toMatchObject({ kind: "thinking" });
});
it("renders status events as system text with optional ctx usage", () => {
expect(
parseAcpxStdoutLine(
emit({ type: "acpx.status", text: "thinking", tag: "agent_thought_chunk" }),
TS,
),
).toEqual([{ kind: "system", ts: TS, text: "thinking" }]);
expect(
parseAcpxStdoutLine(
emit({ type: "acpx.status", tag: "context_window", used: 12000, size: 200000 }),
TS,
),
).toEqual([{ kind: "system", ts: TS, text: "context_window (12000/200000 ctx)" }]);
});
it("emits a tool_call entry that preserves toolCallId, status, and input", () => {
const entries = parseAcpxStdoutLine(
emit({
type: "acpx.tool_call",
name: "read",
toolCallId: "tool-1",
status: "running",
text: "read README.md",
}),
TS,
);
expect(entries).toEqual([
{
kind: "tool_call",
ts: TS,
name: "read",
toolUseId: "tool-1",
input: { text: "read README.md", status: "running" },
},
]);
});
it("emits a paired tool_result entry when a tool_call reports terminal status", () => {
const completed = parseAcpxStdoutLine(
emit({
type: "acpx.tool_call",
name: "read",
toolCallId: "tool-1",
status: "completed",
text: "ok",
}),
TS,
);
expect(completed[1]).toEqual({
kind: "tool_result",
ts: TS,
toolUseId: "tool-1",
toolName: "read",
content: "ok",
isError: false,
});
const failed = parseAcpxStdoutLine(
emit({
type: "acpx.tool_call",
name: "edit",
toolCallId: "tool-2",
status: "failed",
text: "permission denied",
}),
TS,
);
expect(failed[1]).toMatchObject({ kind: "tool_result", isError: true, content: "permission denied" });
});
it("renders acpx.result with summary fallback to stopReason", () => {
const entries = parseAcpxStdoutLine(
emit({ type: "acpx.result", summary: "completed", stopReason: "end_turn" }),
TS,
);
expect(entries[0]).toMatchObject({ kind: "result", text: "completed", subtype: "end_turn", isError: false });
});
it("treats acpx.error as a stderr entry", () => {
const entries = parseAcpxStdoutLine(
emit({ type: "acpx.error", message: "auth required", code: "ACP_AUTH" }),
TS,
);
expect(entries).toEqual([{ kind: "stderr", ts: TS, text: "auth required" }]);
});
it("renders unknown acpx.* events as system entries", () => {
const entries = parseAcpxStdoutLine(
emit({ type: "acpx.misc", message: "unhandled" }),
TS,
);
expect(entries).toEqual([{ kind: "system", ts: TS, text: "unhandled" }]);
});
it("falls back to a stdout entry for non-JSON lines", () => {
const entries = parseAcpxStdoutLine("not json", TS);
expect(entries).toEqual([{ kind: "stdout", ts: TS, text: "not json" }]);
});
});

View File

@@ -0,0 +1,158 @@
import type { TranscriptEntry } from "@paperclipai/adapter-utils";
function parseJson(line: string): Record<string, unknown> | null {
try {
const parsed = JSON.parse(line);
if (typeof parsed !== "object" || parsed === null || Array.isArray(parsed)) return null;
return parsed as Record<string, unknown>;
} catch {
return null;
}
}
function asString(value: unknown, fallback = ""): string {
return typeof value === "string" ? value : fallback;
}
function asNumber(value: unknown, fallback = 0): number {
return typeof value === "number" && Number.isFinite(value) ? value : fallback;
}
function stringify(value: unknown): string {
if (typeof value === "string") return value;
if (value === null || value === undefined) return "";
try {
return JSON.stringify(value, null, 2);
} catch {
return String(value);
}
}
function pickToolUseId(parsed: Record<string, unknown>): string {
return (
asString(parsed.toolCallId) ||
asString(parsed.toolUseId) ||
asString(parsed.id)
);
}
function statusText(parsed: Record<string, unknown>): string {
const text = asString(parsed.text).trim();
const tag = asString(parsed.tag).trim();
const used = asNumber(parsed.used, -1);
const size = asNumber(parsed.size, -1);
const parts: string[] = [];
if (text) parts.push(text);
if (tag && !text) parts.push(tag);
if (used >= 0 && size > 0) parts.push(`(${used}/${size} ctx)`);
return parts.join(" ") || tag || "status";
}
export function parseAcpxStdoutLine(line: string, ts: string): TranscriptEntry[] {
const parsed = parseJson(line);
if (!parsed) return [{ kind: "stdout", ts, text: line }];
const type = asString(parsed.type);
if (type === "acpx.session") {
const agent = asString(parsed.agent, "acpx");
const mode = asString(parsed.mode);
const permissionMode = asString(parsed.permissionMode);
const tail = [mode, permissionMode].filter(Boolean).join(" / ");
return [{
kind: "init",
ts,
model: tail ? `${agent} (${tail})` : agent,
sessionId:
asString(parsed.acpSessionId) ||
asString(parsed.sessionId) ||
asString(parsed.runtimeSessionName),
}];
}
if (type === "acpx.text_delta") {
const text = asString(parsed.text);
if (!text) return [];
const channel = asString(parsed.channel) || asString(parsed.stream);
return [{
kind: channel === "thought" || channel === "thinking" ? "thinking" : "assistant",
ts,
text,
delta: true,
}];
}
if (type === "acpx.tool_call") {
const status = asString(parsed.status);
const text = asString(parsed.text);
const name = asString(parsed.name, "acp_tool");
const toolUseId = pickToolUseId(parsed);
const input =
parsed.input !== undefined
? parsed.input
: text || status
? { ...(text ? { text } : {}), ...(status ? { status } : {}) }
: {};
const entries: TranscriptEntry[] = [
{
kind: "tool_call",
ts,
name,
toolUseId: toolUseId || undefined,
input,
},
];
if (status === "completed" || status === "failed" || status === "cancelled") {
entries.push({
kind: "tool_result",
ts,
toolUseId: toolUseId || name,
toolName: name,
content: text || status,
isError: status !== "completed",
});
}
return entries;
}
if (type === "acpx.tool_result") {
return [{
kind: "tool_result",
ts,
toolUseId: pickToolUseId(parsed) || asString(parsed.name, "acp_tool"),
toolName: asString(parsed.name) || undefined,
content: stringify(parsed.content ?? parsed.output ?? parsed.error),
isError: parsed.isError === true || parsed.error !== undefined,
}];
}
if (type === "acpx.status") {
return [{ kind: "system", ts, text: statusText(parsed) }];
}
if (type === "acpx.result") {
return [{
kind: "result",
ts,
text: asString(parsed.summary, asString(parsed.stopReason, asString(parsed.text))),
inputTokens: asNumber(parsed.inputTokens),
outputTokens: asNumber(parsed.outputTokens),
cachedTokens: asNumber(parsed.cachedTokens),
costUsd: asNumber(parsed.costUsd),
subtype: asString(parsed.subtype, asString(parsed.stopReason, "acpx.result")),
isError: parsed.isError === true,
errors: Array.isArray(parsed.errors)
? parsed.errors.map((error) => stringify(error)).filter(Boolean)
: [],
}];
}
if (type === "acpx.error") {
return [{ kind: "stderr", ts, text: asString(parsed.message, line) }];
}
if (type.startsWith("acpx.")) {
return [{ kind: "system", ts, text: asString(parsed.message, type) }];
}
return [{ kind: "stdout", ts, text: line }];
}

View File

@@ -0,0 +1,8 @@
{
"extends": "../../../tsconfig.base.json",
"compilerOptions": {
"outDir": "dist",
"rootDir": "src"
},
"include": ["src"]
}

View File

@@ -0,0 +1,7 @@
import { defineConfig } from "vitest/config";
export default defineConfig({
test: {
environment: "node",
},
});

View File

@@ -1,3 +1,5 @@
import type { AdapterModelProfileDefinition } from "@paperclipai/adapter-utils";
export const type = "claude_local";
export const label = "Claude Code (local)";
@@ -10,6 +12,19 @@ export const models = [
{ id: "claude-haiku-4-5-20251001", label: "Claude Haiku 4.5" },
];
export const modelProfiles: AdapterModelProfileDefinition[] = [
{
key: "cheap",
label: "Cheap",
description: "Use Claude Sonnet as the lower-cost Claude Code lane while preserving the agent's primary model.",
adapterConfig: {
model: "claude-sonnet-4-6",
effort: "low",
},
source: "adapter_default",
},
];
export const agentConfigurationDoc = `# claude_local agent configuration
Adapter: claude_local

View File

@@ -10,6 +10,7 @@ const {
prepareWorkspaceForSshExecution,
restoreWorkspaceFromSshExecution,
syncDirectoryToSsh,
startAdapterExecutionTargetPaperclipBridge,
} = vi.hoisted(() => ({
runChildProcess: vi.fn(async () => ({
exitCode: 0,
@@ -29,6 +30,14 @@ const {
prepareWorkspaceForSshExecution: vi.fn(async () => undefined),
restoreWorkspaceFromSshExecution: vi.fn(async () => undefined),
syncDirectoryToSsh: vi.fn(async () => undefined),
startAdapterExecutionTargetPaperclipBridge: vi.fn(async () => ({
env: {
PAPERCLIP_API_URL: "http://127.0.0.1:4310",
PAPERCLIP_API_KEY: "bridge-token",
PAPERCLIP_API_BRIDGE_MODE: "queue_v1",
},
stop: async () => {},
})),
}));
vi.mock("@paperclipai/adapter-utils/server-utils", async () => {
@@ -55,6 +64,16 @@ vi.mock("@paperclipai/adapter-utils/ssh", async () => {
};
});
vi.mock("@paperclipai/adapter-utils/execution-target", async () => {
const actual = await vi.importActual<typeof import("@paperclipai/adapter-utils/execution-target")>(
"@paperclipai/adapter-utils/execution-target",
);
return {
...actual,
startAdapterExecutionTargetPaperclipBridge,
};
});
import { execute } from "./execute.js";
describe("claude remote execution", () => {
@@ -73,8 +92,10 @@ describe("claude remote execution", () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-claude-remote-"));
cleanupDirs.push(rootDir);
const workspaceDir = path.join(rootDir, "workspace");
const alternateWorkspaceDir = path.join(rootDir, "workspace-other");
const instructionsPath = path.join(rootDir, "instructions.md");
await mkdir(workspaceDir, { recursive: true });
await mkdir(alternateWorkspaceDir, { recursive: true });
await writeFile(instructionsPath, "Use the remote workspace.\n", "utf8");
await execute({
@@ -100,7 +121,27 @@ describe("claude remote execution", () => {
paperclipWorkspace: {
cwd: workspaceDir,
source: "project_primary",
strategy: "git_worktree",
workspaceId: "workspace-1",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
branchName: "feature/remote-claude",
worktreePath: workspaceDir,
},
paperclipWorkspaces: [
{
workspaceId: "workspace-1",
cwd: workspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
cwd: alternateWorkspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
],
},
executionTransport: {
remoteExecution: {
@@ -112,7 +153,6 @@ describe("claude remote execution", () => {
privateKey: "PRIVATE KEY",
knownHosts: "[127.0.0.1]:2222 ssh-ed25519 AAAA",
strictHostKeyChecking: true,
paperclipApiUrl: "http://198.51.100.10:3102",
},
},
onLog: async () => {},
@@ -136,8 +176,25 @@ describe("claude remote execution", () => {
expect(call?.[2]).toContain("/remote/workspace/.paperclip-runtime/claude/skills/agent-instructions.md");
expect(call?.[2]).toContain("--add-dir");
expect(call?.[2]).toContain("/remote/workspace/.paperclip-runtime/claude/skills");
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://198.51.100.10:3102");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe("/remote/workspace");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_WORKTREE_PATH).toBeUndefined();
expect(JSON.parse(call?.[3].env.PAPERCLIP_WORKSPACES_JSON ?? "[]")).toEqual([
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
]);
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:4310");
expect(call?.[3].env.PAPERCLIP_API_BRIDGE_MODE).toBe("queue_v1");
expect(call?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
expect(startAdapterExecutionTargetPaperclipBridge).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledWith(expect.objectContaining({
localDir: workspaceDir,

View File

@@ -5,7 +5,6 @@ import type { AdapterExecutionContext, AdapterExecutionResult } from "@paperclip
import type { RunProcessResult } from "@paperclipai/adapter-utils/server-utils";
import {
adapterExecutionTargetIsRemote,
adapterExecutionTargetPaperclipApiUrl,
adapterExecutionTargetRemoteCwd,
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetSessionMatches,
@@ -13,6 +12,7 @@ import {
adapterExecutionTargetUsesPaperclipBridge,
describeAdapterExecutionTarget,
ensureAdapterExecutionTargetCommandResolvable,
ensureAdapterExecutionTargetRuntimeCommandInstalled,
prepareAdapterExecutionTargetRuntime,
readAdapterExecutionTarget,
resolveAdapterExecutionTargetCommandForLogs,
@@ -36,6 +36,7 @@ import {
ensurePathInEnv,
renderTemplate,
renderPaperclipWakePrompt,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
} from "@paperclipai/adapter-utils/server-utils";
@@ -61,8 +62,10 @@ interface ClaudeExecutionInput {
agent: AdapterExecutionContext["agent"];
config: Record<string, unknown>;
context: Record<string, unknown>;
runtimeCommandSpec?: AdapterExecutionContext["runtimeCommandSpec"];
executionTarget?: ReturnType<typeof readAdapterExecutionTarget>;
authToken?: string;
onLog?: (stream: "stdout" | "stderr", chunk: string) => Promise<void>;
}
interface ClaudeRuntimeConfig {
@@ -112,7 +115,8 @@ function resolveClaudeBillingType(env: Record<string, string>): "api" | "subscri
}
async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<ClaudeRuntimeConfig> {
const { runId, agent, config, context, executionTarget, authToken } = input;
const { runId, agent, config, context, runtimeCommandSpec, executionTarget, authToken } = input;
const onLog = input.onLog ?? (async () => {});
const command = asString(config.command, "claude");
const workspaceContext = parseObject(context.paperclipWorkspace);
@@ -145,6 +149,15 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
const useConfiguredInsteadOfAgentHome = workspaceSource === "agent_home" && configuredCwd.length > 0;
const effectiveWorkspaceCwd = useConfiguredInsteadOfAgentHome ? "" : workspaceCwd;
const cwd = effectiveWorkspaceCwd || configuredCwd || process.cwd();
const executionTargetIsRemote = adapterExecutionTargetIsRemote(executionTarget);
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const shapedWorkspaceEnv = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: effectiveWorkspaceCwd,
workspaceWorktreePath,
workspaceHints,
executionTargetIsRemote,
executionCwd: effectiveExecutionCwd,
});
await ensureAbsoluteDirectory(cwd, { createIfMissing: true });
const envConfig = parseObject(config.env);
@@ -200,18 +213,18 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
env.PAPERCLIP_WAKE_PAYLOAD_JSON = wakePayloadJson;
}
applyPaperclipWorkspaceEnv(env, {
workspaceCwd: effectiveWorkspaceCwd,
workspaceCwd: shapedWorkspaceEnv.workspaceCwd,
workspaceSource,
workspaceStrategy,
workspaceId,
workspaceRepoUrl,
workspaceRepoRef,
workspaceBranch,
workspaceWorktreePath,
workspaceWorktreePath: shapedWorkspaceEnv.workspaceWorktreePath,
agentHome,
});
if (workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(workspaceHints);
if (shapedWorkspaceEnv.workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(shapedWorkspaceEnv.workspaceHints);
}
if (runtimeServiceIntents.length > 0) {
env.PAPERCLIP_RUNTIME_SERVICE_INTENTS_JSON = JSON.stringify(runtimeServiceIntents);
@@ -222,11 +235,6 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
if (runtimePrimaryUrl) {
env.PAPERCLIP_RUNTIME_PRIMARY_URL = runtimePrimaryUrl;
}
const targetPaperclipApiUrl = adapterExecutionTargetPaperclipApiUrl(executionTarget);
if (targetPaperclipApiUrl) {
env.PAPERCLIP_API_URL = targetPaperclipApiUrl;
}
for (const [key, value] of Object.entries(envConfig)) {
if (typeof value === "string") env[key] = value;
}
@@ -235,7 +243,24 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
env.PAPERCLIP_API_KEY = authToken;
}
const runtimeEnv = ensurePathInEnv({ ...process.env, ...env });
const runtimeEnv = Object.fromEntries(
Object.entries(ensurePathInEnv({ ...process.env, ...env })).filter(
(entry): entry is [string, string] => typeof entry[1] === "string",
),
);
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId,
target: executionTarget,
installCommand: runtimeCommandSpec?.installCommand,
detectCommand: runtimeCommandSpec?.detectCommand,
cwd,
env: runtimeEnv,
timeoutSec,
graceSec,
onLog,
});
await ensureAdapterExecutionTargetCommandResolvable(command, executionTarget, cwd, runtimeEnv);
const resolvedCommand = await resolveAdapterExecutionTargetCommandForLogs(command, executionTarget, cwd, runtimeEnv);
const loggedEnv = buildInvocationEnvForLogs(env, {
@@ -244,8 +269,6 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
resolvedCommand,
});
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
const extraArgs = (() => {
const fromExtraArgs = asStringArray(config.extraArgs);
if (fromExtraArgs.length > 0) return fromExtraArgs;
@@ -331,8 +354,10 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
agent,
config,
context,
runtimeCommandSpec: ctx.runtimeCommandSpec,
executionTarget,
authToken,
onLog,
});
const {
command,
@@ -777,6 +802,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
const transientUpstream =
failed &&
!loginMeta.requiresLogin &&
!clearSessionForMaxTurns &&
isClaudeTransientUpstreamError({
parsed,
stdout: proc.stdout,
@@ -793,11 +819,14 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
: null;
const resolvedErrorCode = loginMeta.requiresLogin
? "claude_auth_required"
: failed && clearSessionForMaxTurns
? "max_turns_exhausted"
: transientUpstream
? "claude_transient_upstream"
: null;
const mergedResultJson: Record<string, unknown> = {
...parsed,
...(failed && clearSessionForMaxTurns ? { stopReason: "max_turns_exhausted" } : {}),
...(transientUpstream ? { errorFamily: "transient_upstream" } : {}),
...(transientRetryNotBefore ? { retryNotBefore: transientRetryNotBefore.toISOString() } : {}),
...(transientRetryNotBefore ? { transientRetryNotBefore: transientRetryNotBefore.toISOString() } : {}),

View File

@@ -170,11 +170,19 @@ export function isClaudeMaxTurnsResult(parsed: Record<string, unknown> | null |
const subtype = asString(parsed.subtype, "").trim().toLowerCase();
if (subtype === "error_max_turns") return true;
const stopReason = asString(parsed.stop_reason, "").trim().toLowerCase();
if (stopReason === "max_turns") return true;
const structuredStopReasons = [
parsed.stop_reason,
parsed.stopReason,
parsed.error_code,
parsed.errorCode,
].map((value) => asString(value, "").trim().toLowerCase());
const resultText = asString(parsed.result, "").trim();
return /max(?:imum)?\s+turns?/i.test(resultText);
return structuredStopReasons.some((reason) =>
reason === "max_turns" ||
reason === "max_turns_exhausted" ||
reason === "turn_limit" ||
reason === "turn_limit_exhausted",
);
}
export function isClaudeUnknownSessionError(parsed: Record<string, unknown>): boolean {

View File

@@ -66,8 +66,6 @@ export function buildClaudeLocalConfig(v: CreateConfigValues): Record<string, un
const ac: Record<string, unknown> = {};
if (v.cwd) ac.cwd = v.cwd;
if (v.instructionsFilePath) ac.instructionsFilePath = v.instructionsFilePath;
if (v.promptTemplate) ac.promptTemplate = v.promptTemplate;
if (v.bootstrapPrompt) ac.bootstrapPromptTemplate = v.bootstrapPrompt;
if (v.model) ac.model = v.model;
if (v.thinkingEffort) ac.effort = v.thinkingEffort;
if (v.chrome) ac.chrome = true;

View File

@@ -1,5 +1,8 @@
import type { AdapterModelProfileDefinition } from "@paperclipai/adapter-utils";
export const type = "codex_local";
export const label = "Codex (local)";
export const DEFAULT_CODEX_LOCAL_MODEL = "gpt-5.3-codex";
export const DEFAULT_CODEX_LOCAL_BYPASS_APPROVALS_AND_SANDBOX = true;
export const CODEX_LOCAL_FAST_MODE_SUPPORTED_MODELS = ["gpt-5.4"] as const;
@@ -40,6 +43,19 @@ export const models = [
{ id: "codex-mini-latest", label: "Codex Mini" },
];
export const modelProfiles: AdapterModelProfileDefinition[] = [
{
key: "cheap",
label: "Cheap",
description: "Use the lowest-cost known Codex local model lane without changing the primary model.",
adapterConfig: {
model: "gpt-5.3-codex-spark",
modelReasoningEffort: "low",
},
source: "adapter_default",
},
];
export const agentConfigurationDoc = `# codex_local agent configuration
Adapter: codex_local

View File

@@ -10,6 +10,7 @@ const {
prepareWorkspaceForSshExecution,
restoreWorkspaceFromSshExecution,
syncDirectoryToSsh,
startAdapterExecutionTargetPaperclipBridge,
} = vi.hoisted(() => ({
runChildProcess: vi.fn(async () => ({
exitCode: 1,
@@ -25,6 +26,14 @@ const {
prepareWorkspaceForSshExecution: vi.fn(async () => undefined),
restoreWorkspaceFromSshExecution: vi.fn(async () => undefined),
syncDirectoryToSsh: vi.fn(async () => undefined),
startAdapterExecutionTargetPaperclipBridge: vi.fn(async () => ({
env: {
PAPERCLIP_API_URL: "http://127.0.0.1:4310",
PAPERCLIP_API_KEY: "bridge-token",
PAPERCLIP_API_BRIDGE_MODE: "queue_v1",
},
stop: async () => {},
})),
}));
vi.mock("@paperclipai/adapter-utils/server-utils", async () => {
@@ -51,6 +60,16 @@ vi.mock("@paperclipai/adapter-utils/ssh", async () => {
};
});
vi.mock("@paperclipai/adapter-utils/execution-target", async () => {
const actual = await vi.importActual<typeof import("@paperclipai/adapter-utils/execution-target")>(
"@paperclipai/adapter-utils/execution-target",
);
return {
...actual,
startAdapterExecutionTargetPaperclipBridge,
};
});
import { execute } from "./execute.js";
describe("codex remote execution", () => {
@@ -74,6 +93,8 @@ describe("codex remote execution", () => {
await mkdir(codexHomeDir, { recursive: true });
await writeFile(path.join(rootDir, "instructions.md"), "Use the remote workspace.\n", "utf8");
await writeFile(path.join(codexHomeDir, "auth.json"), "{}", "utf8");
const alternateWorkspaceDir = path.join(rootDir, "alternate-workspace");
await mkdir(alternateWorkspaceDir, { recursive: true });
await execute({
runId: "run-1",
@@ -100,7 +121,27 @@ describe("codex remote execution", () => {
paperclipWorkspace: {
cwd: workspaceDir,
source: "project_primary",
strategy: "git_worktree",
workspaceId: "workspace-1",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
branchName: "feature/remote-codex",
worktreePath: workspaceDir,
},
paperclipWorkspaces: [
{
workspaceId: "workspace-1",
cwd: workspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
cwd: alternateWorkspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
],
},
executionTransport: {
remoteExecution: {
@@ -134,7 +175,25 @@ describe("codex remote execution", () => {
| [string, string, string[], { env: Record<string, string>; remoteExecution?: { remoteCwd: string } | null }]
| undefined;
expect(call?.[3].env.CODEX_HOME).toBe("/remote/workspace/.paperclip-runtime/codex/home");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe("/remote/workspace");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_WORKTREE_PATH).toBeUndefined();
expect(JSON.parse(call?.[3].env.PAPERCLIP_WORKSPACES_JSON ?? "[]")).toEqual([
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
]);
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:4310");
expect(call?.[3].env.PAPERCLIP_API_BRIDGE_MODE).toBe("queue_v1");
expect(call?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
expect(startAdapterExecutionTargetPaperclipBridge).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledWith(expect.objectContaining({
localDir: workspaceDir,

View File

@@ -4,13 +4,13 @@ import { fileURLToPath } from "node:url";
import { inferOpenAiCompatibleBiller, type AdapterExecutionContext, type AdapterExecutionResult } from "@paperclipai/adapter-utils";
import {
adapterExecutionTargetIsRemote,
adapterExecutionTargetPaperclipApiUrl,
adapterExecutionTargetRemoteCwd,
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetSessionMatches,
adapterExecutionTargetUsesPaperclipBridge,
describeAdapterExecutionTarget,
ensureAdapterExecutionTargetCommandResolvable,
ensureAdapterExecutionTargetRuntimeCommandInstalled,
prepareAdapterExecutionTargetRuntime,
readAdapterExecutionTarget,
resolveAdapterExecutionTargetCommandForLogs,
@@ -31,6 +31,7 @@ import {
resolvePaperclipDesiredSkillNames,
renderTemplate,
renderPaperclipWakePrompt,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
joinPromptSections,
@@ -348,6 +349,13 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
},
);
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const shapedWorkspaceEnv = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: effectiveWorkspaceCwd,
workspaceWorktreePath,
workspaceHints,
executionTargetIsRemote,
executionCwd: effectiveExecutionCwd,
});
const preparedExecutionTargetRuntime = executionTargetIsRemote
? await (async () => {
await onLog(
@@ -426,18 +434,18 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
env.PAPERCLIP_WAKE_PAYLOAD_JSON = wakePayloadJson;
}
applyPaperclipWorkspaceEnv(env, {
workspaceCwd: effectiveWorkspaceCwd,
workspaceCwd: shapedWorkspaceEnv.workspaceCwd,
workspaceSource,
workspaceStrategy,
workspaceId,
workspaceRepoUrl,
workspaceRepoRef,
workspaceBranch,
workspaceWorktreePath,
workspaceWorktreePath: shapedWorkspaceEnv.workspaceWorktreePath,
agentHome,
});
if (workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(workspaceHints);
if (shapedWorkspaceEnv.workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(shapedWorkspaceEnv.workspaceHints);
}
if (runtimeServiceIntents.length > 0) {
env.PAPERCLIP_RUNTIME_SERVICE_INTENTS_JSON = JSON.stringify(runtimeServiceIntents);
@@ -448,10 +456,6 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
if (runtimePrimaryUrl) {
env.PAPERCLIP_RUNTIME_PRIMARY_URL = runtimePrimaryUrl;
}
const targetPaperclipApiUrl = adapterExecutionTargetPaperclipApiUrl(executionTarget);
if (targetPaperclipApiUrl) {
env.PAPERCLIP_API_URL = targetPaperclipApiUrl;
}
for (const [k, v] of Object.entries(envConfig)) {
if (typeof v === "string") env[k] = v;
}
@@ -478,7 +482,22 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
),
);
const billingType = resolveCodexBillingType(effectiveEnv);
const runtimeEnv = ensurePathInEnv(effectiveEnv);
const runtimeEnv = Object.fromEntries(
Object.entries(ensurePathInEnv(effectiveEnv)).filter(
(entry): entry is [string, string] => typeof entry[1] === "string",
),
);
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId,
target: executionTarget,
installCommand: ctx.runtimeCommandSpec?.installCommand,
detectCommand: ctx.runtimeCommandSpec?.detectCommand,
cwd,
env: runtimeEnv,
timeoutSec: asNumber(config.timeoutSec, 0),
graceSec: asNumber(config.graceSec, 20),
onLog,
});
await ensureAdapterExecutionTargetCommandResolvable(command, executionTarget, cwd, runtimeEnv);
const resolvedCommand = await resolveAdapterExecutionTargetCommandForLogs(command, executionTarget, cwd, runtimeEnv);
const loggedEnv = buildInvocationEnvForLogs(env, {

View File

@@ -70,8 +70,6 @@ export function buildCodexLocalConfig(v: CreateConfigValues): Record<string, unk
const ac: Record<string, unknown> = {};
if (v.cwd) ac.cwd = v.cwd;
if (v.instructionsFilePath) ac.instructionsFilePath = v.instructionsFilePath;
if (v.promptTemplate) ac.promptTemplate = v.promptTemplate;
if (v.bootstrapPrompt) ac.bootstrapPromptTemplate = v.bootstrapPrompt;
ac.model = v.model || DEFAULT_CODEX_LOCAL_MODEL;
if (v.thinkingEffort) ac.modelReasoningEffort = v.thinkingEffort;
ac.timeoutSec = 0;

View File

@@ -1,5 +1,8 @@
import type { AdapterModelProfileDefinition } from "@paperclipai/adapter-utils";
export const type = "cursor";
export const label = "Cursor CLI (local)";
export const DEFAULT_CURSOR_LOCAL_MODEL = "auto";
const CURSOR_FALLBACK_MODEL_IDS = [
@@ -46,6 +49,18 @@ const CURSOR_FALLBACK_MODEL_IDS = [
export const models = CURSOR_FALLBACK_MODEL_IDS.map((id) => ({ id, label: id }));
export const modelProfiles: AdapterModelProfileDefinition[] = [
{
key: "cheap",
label: "Cheap",
description: "Use Cursor's known Codex mini model as the budget lane instead of assuming auto is cheap.",
adapterConfig: {
model: "gpt-5.1-codex-mini",
},
source: "adapter_default",
},
];
export const agentConfigurationDoc = `# cursor agent configuration
Adapter: cursor

View File

@@ -11,6 +11,7 @@ const {
restoreWorkspaceFromSshExecution,
runSshCommand,
syncDirectoryToSsh,
startAdapterExecutionTargetPaperclipBridge,
} = vi.hoisted(() => ({
runChildProcess: vi.fn(async () => ({
exitCode: 0,
@@ -35,6 +36,14 @@ const {
exitCode: 0,
})),
syncDirectoryToSsh: vi.fn(async () => undefined),
startAdapterExecutionTargetPaperclipBridge: vi.fn(async () => ({
env: {
PAPERCLIP_API_URL: "http://127.0.0.1:4310",
PAPERCLIP_API_KEY: "bridge-token",
PAPERCLIP_API_BRIDGE_MODE: "queue_v1",
},
stop: async () => {},
})),
}));
vi.mock("@paperclipai/adapter-utils/server-utils", async () => {
@@ -62,6 +71,16 @@ vi.mock("@paperclipai/adapter-utils/ssh", async () => {
};
});
vi.mock("@paperclipai/adapter-utils/execution-target", async () => {
const actual = await vi.importActual<typeof import("@paperclipai/adapter-utils/execution-target")>(
"@paperclipai/adapter-utils/execution-target",
);
return {
...actual,
startAdapterExecutionTargetPaperclipBridge,
};
});
import { execute } from "./execute.js";
describe("cursor remote execution", () => {
@@ -80,7 +99,9 @@ describe("cursor remote execution", () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-cursor-remote-"));
cleanupDirs.push(rootDir);
const workspaceDir = path.join(rootDir, "workspace");
const alternateWorkspaceDir = path.join(rootDir, "workspace-other");
await mkdir(workspaceDir, { recursive: true });
await mkdir(alternateWorkspaceDir, { recursive: true });
const result = await execute({
runId: "run-1",
@@ -105,6 +126,20 @@ describe("cursor remote execution", () => {
cwd: workspaceDir,
source: "project_primary",
},
paperclipWorkspaces: [
{
workspaceId: "workspace-1",
cwd: workspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
cwd: alternateWorkspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
],
},
executionTransport: {
remoteExecution: {
@@ -116,7 +151,6 @@ describe("cursor remote execution", () => {
privateKey: "PRIVATE KEY",
knownHosts: "[127.0.0.1]:2222 ssh-ed25519 AAAA",
strictHostKeyChecking: true,
paperclipApiUrl: "http://198.51.100.10:3102",
},
},
onLog: async () => {},
@@ -131,7 +165,6 @@ describe("cursor remote execution", () => {
port: 2222,
username: "fixture",
remoteCwd: "/remote/workspace",
paperclipApiUrl: "http://198.51.100.10:3102",
},
});
expect(prepareWorkspaceForSshExecution).toHaveBeenCalledTimes(1);
@@ -150,8 +183,24 @@ describe("cursor remote execution", () => {
| undefined;
expect(call?.[2]).toContain("--workspace");
expect(call?.[2]).toContain("/remote/workspace");
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://198.51.100.10:3102");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe("/remote/workspace");
expect(JSON.parse(call?.[3].env.PAPERCLIP_WORKSPACES_JSON ?? "[]")).toEqual([
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
]);
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:4310");
expect(call?.[3].env.PAPERCLIP_API_BRIDGE_MODE).toBe("queue_v1");
expect(call?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
expect(startAdapterExecutionTargetPaperclipBridge).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledTimes(1);
});

View File

@@ -5,7 +5,6 @@ import { fileURLToPath } from "node:url";
import { inferOpenAiCompatibleBiller, type AdapterExecutionContext, type AdapterExecutionResult } from "@paperclipai/adapter-utils";
import {
adapterExecutionTargetIsRemote,
adapterExecutionTargetPaperclipApiUrl,
adapterExecutionTargetRemoteCwd,
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetSessionMatches,
@@ -13,6 +12,7 @@ import {
adapterExecutionTargetUsesPaperclipBridge,
describeAdapterExecutionTarget,
ensureAdapterExecutionTargetCommandResolvable,
ensureAdapterExecutionTargetRuntimeCommandInstalled,
prepareAdapterExecutionTargetRuntime,
readAdapterExecutionTarget,
readAdapterExecutionTargetHomeDir,
@@ -37,6 +37,7 @@ import {
removeMaintainerOnlySkillSymlinks,
renderTemplate,
renderPaperclipWakePrompt,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
joinPromptSections,
@@ -222,6 +223,13 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
const useConfiguredInsteadOfAgentHome = workspaceSource === "agent_home" && configuredCwd.length > 0;
const effectiveWorkspaceCwd = useConfiguredInsteadOfAgentHome ? "" : workspaceCwd;
const cwd = effectiveWorkspaceCwd || configuredCwd || process.cwd();
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const shapedWorkspaceEnv = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: effectiveWorkspaceCwd,
workspaceHints,
executionTargetIsRemote,
executionCwd: effectiveExecutionCwd,
});
await ensureAbsoluteDirectory(cwd, { createIfMissing: true });
const cursorSkillEntries = await readPaperclipRuntimeSkillEntries(config, __moduleDir);
const desiredCursorSkillNames = resolvePaperclipDesiredSkillNames(config, cursorSkillEntries);
@@ -282,19 +290,15 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
env.PAPERCLIP_WAKE_PAYLOAD_JSON = wakePayloadJson;
}
applyPaperclipWorkspaceEnv(env, {
workspaceCwd: effectiveWorkspaceCwd,
workspaceCwd: shapedWorkspaceEnv.workspaceCwd,
workspaceSource,
workspaceId,
workspaceRepoUrl,
workspaceRepoRef,
agentHome,
});
if (workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(workspaceHints);
}
const targetPaperclipApiUrl = adapterExecutionTargetPaperclipApiUrl(executionTarget);
if (targetPaperclipApiUrl) {
env.PAPERCLIP_API_URL = targetPaperclipApiUrl;
if (shapedWorkspaceEnv.workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(shapedWorkspaceEnv.workspaceHints);
}
for (const [k, v] of Object.entries(envConfig)) {
if (typeof v === "string") env[k] = v;
@@ -304,6 +308,17 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
}
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId,
target: executionTarget,
installCommand: ctx.runtimeCommandSpec?.installCommand,
detectCommand: ctx.runtimeCommandSpec?.detectCommand,
cwd,
env,
timeoutSec,
graceSec,
onLog,
});
// Probe the sandbox before the managed-home override so we discover
// cursor-agent from the real system HOME (e.g. ~/.local/bin/cursor-agent).
// The managed HOME set later is for runtime isolation, not for finding the CLI.
@@ -339,7 +354,6 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
return asStringArray(config.args);
})();
const autoTrustEnabled = !hasCursorTrustBypassArg(extraArgs);
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
let restoreRemoteWorkspace: (() => Promise<void>) | null = null;
let localSkillsDir: string | null = null;
let remoteRuntimeRootDir: string | null = null;

View File

@@ -61,8 +61,6 @@ export function buildCursorLocalConfig(v: CreateConfigValues): Record<string, un
const ac: Record<string, unknown> = {};
if (v.cwd) ac.cwd = v.cwd;
if (v.instructionsFilePath) ac.instructionsFilePath = v.instructionsFilePath;
if (v.promptTemplate) ac.promptTemplate = v.promptTemplate;
if (v.bootstrapPrompt) ac.bootstrapPromptTemplate = v.bootstrapPrompt;
ac.model = v.model || DEFAULT_CURSOR_LOCAL_MODEL;
const mode = normalizeMode(v.thinkingEffort);
if (mode) ac.mode = mode;

View File

@@ -1,5 +1,8 @@
import type { AdapterModelProfileDefinition } from "@paperclipai/adapter-utils";
export const type = "gemini_local";
export const label = "Gemini CLI (local)";
export const DEFAULT_GEMINI_LOCAL_MODEL = "auto";
export const models = [
@@ -11,6 +14,18 @@ export const models = [
{ id: "gemini-2.0-flash-lite", label: "Gemini 2.0 Flash Lite" },
];
export const modelProfiles: AdapterModelProfileDefinition[] = [
{
key: "cheap",
label: "Cheap",
description: "Use Gemini Flash Lite as the budget Gemini CLI lane while preserving the primary model.",
adapterConfig: {
model: "gemini-2.5-flash-lite",
},
source: "adapter_default",
},
];
export const agentConfigurationDoc = `# gemini_local agent configuration
Adapter: gemini_local

View File

@@ -11,6 +11,7 @@ const {
restoreWorkspaceFromSshExecution,
runSshCommand,
syncDirectoryToSsh,
startAdapterExecutionTargetPaperclipBridge,
} = vi.hoisted(() => ({
runChildProcess: vi.fn(async () => ({
exitCode: 0,
@@ -41,6 +42,14 @@ const {
exitCode: 0,
})),
syncDirectoryToSsh: vi.fn(async () => undefined),
startAdapterExecutionTargetPaperclipBridge: vi.fn(async () => ({
env: {
PAPERCLIP_API_URL: "http://127.0.0.1:4310",
PAPERCLIP_API_KEY: "bridge-token",
PAPERCLIP_API_BRIDGE_MODE: "queue_v1",
},
stop: async () => {},
})),
}));
vi.mock("@paperclipai/adapter-utils/server-utils", async () => {
@@ -68,6 +77,16 @@ vi.mock("@paperclipai/adapter-utils/ssh", async () => {
};
});
vi.mock("@paperclipai/adapter-utils/execution-target", async () => {
const actual = await vi.importActual<typeof import("@paperclipai/adapter-utils/execution-target")>(
"@paperclipai/adapter-utils/execution-target",
);
return {
...actual,
startAdapterExecutionTargetPaperclipBridge,
};
});
import { execute } from "./execute.js";
describe("gemini remote execution", () => {
@@ -86,7 +105,9 @@ describe("gemini remote execution", () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-gemini-remote-"));
cleanupDirs.push(rootDir);
const workspaceDir = path.join(rootDir, "workspace");
const alternateWorkspaceDir = path.join(rootDir, "workspace-other");
await mkdir(workspaceDir, { recursive: true });
await mkdir(alternateWorkspaceDir, { recursive: true });
const result = await execute({
runId: "run-1",
@@ -111,6 +132,20 @@ describe("gemini remote execution", () => {
cwd: workspaceDir,
source: "project_primary",
},
paperclipWorkspaces: [
{
workspaceId: "workspace-1",
cwd: workspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
cwd: alternateWorkspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
],
},
executionTransport: {
remoteExecution: {
@@ -122,7 +157,6 @@ describe("gemini remote execution", () => {
privateKey: "PRIVATE KEY",
knownHosts: "[127.0.0.1]:2222 ssh-ed25519 AAAA",
strictHostKeyChecking: true,
paperclipApiUrl: "http://198.51.100.10:3102",
},
},
onLog: async () => {},
@@ -137,7 +171,6 @@ describe("gemini remote execution", () => {
port: 2222,
username: "fixture",
remoteCwd: "/remote/workspace",
paperclipApiUrl: "http://198.51.100.10:3102",
},
});
expect(prepareWorkspaceForSshExecution).toHaveBeenCalledTimes(1);
@@ -154,8 +187,24 @@ describe("gemini remote execution", () => {
const call = runChildProcess.mock.calls[0] as unknown as
| [string, string, string[], { env: Record<string, string>; remoteExecution?: { remoteCwd: string } | null }]
| undefined;
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://198.51.100.10:3102");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe("/remote/workspace");
expect(JSON.parse(call?.[3].env.PAPERCLIP_WORKSPACES_JSON ?? "[]")).toEqual([
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
]);
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:4310");
expect(call?.[3].env.PAPERCLIP_API_BRIDGE_MODE).toBe("queue_v1");
expect(call?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
expect(startAdapterExecutionTargetPaperclipBridge).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledTimes(1);
});

View File

@@ -6,7 +6,6 @@ import { fileURLToPath } from "node:url";
import type { AdapterExecutionContext, AdapterExecutionResult } from "@paperclipai/adapter-utils";
import {
adapterExecutionTargetIsRemote,
adapterExecutionTargetPaperclipApiUrl,
adapterExecutionTargetRemoteCwd,
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetSessionMatches,
@@ -14,6 +13,7 @@ import {
adapterExecutionTargetUsesPaperclipBridge,
describeAdapterExecutionTarget,
ensureAdapterExecutionTargetCommandResolvable,
ensureAdapterExecutionTargetRuntimeCommandInstalled,
prepareAdapterExecutionTargetRuntime,
readAdapterExecutionTarget,
readAdapterExecutionTargetHomeDir,
@@ -40,6 +40,7 @@ import {
parseObject,
renderTemplate,
renderPaperclipWakePrompt,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
runChildProcess,
@@ -200,6 +201,13 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
const useConfiguredInsteadOfAgentHome = workspaceSource === "agent_home" && configuredCwd.length > 0;
const effectiveWorkspaceCwd = useConfiguredInsteadOfAgentHome ? "" : workspaceCwd;
const cwd = effectiveWorkspaceCwd || configuredCwd || process.cwd();
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const shapedWorkspaceEnv = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: effectiveWorkspaceCwd,
workspaceHints,
executionTargetIsRemote,
executionCwd: effectiveExecutionCwd,
});
await ensureAbsoluteDirectory(cwd, { createIfMissing: true });
const geminiSkillEntries = await readPaperclipRuntimeSkillEntries(config, __moduleDir);
const desiredGeminiSkillNames = resolvePaperclipDesiredSkillNames(config, geminiSkillEntries);
@@ -244,17 +252,16 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
if (linkedIssueIds.length > 0) env.PAPERCLIP_LINKED_ISSUE_IDS = linkedIssueIds.join(",");
if (wakePayloadJson) env.PAPERCLIP_WAKE_PAYLOAD_JSON = wakePayloadJson;
applyPaperclipWorkspaceEnv(env, {
workspaceCwd: effectiveWorkspaceCwd,
workspaceCwd: shapedWorkspaceEnv.workspaceCwd,
workspaceSource,
workspaceId,
workspaceRepoUrl,
workspaceRepoRef,
agentHome,
});
if (workspaceHints.length > 0) env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(workspaceHints);
const targetPaperclipApiUrl = adapterExecutionTargetPaperclipApiUrl(executionTarget);
if (targetPaperclipApiUrl) env.PAPERCLIP_API_URL = targetPaperclipApiUrl;
if (shapedWorkspaceEnv.workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(shapedWorkspaceEnv.workspaceHints);
}
for (const [key, value] of Object.entries(envConfig)) {
if (typeof value === "string") env[key] = value;
}
@@ -267,7 +274,24 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
),
);
const billingType = resolveGeminiBillingType(effectiveEnv);
const runtimeEnv = ensurePathInEnv(effectiveEnv);
const runtimeEnv = Object.fromEntries(
Object.entries(ensurePathInEnv(effectiveEnv)).filter(
(entry): entry is [string, string] => typeof entry[1] === "string",
),
);
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId,
target: executionTarget,
installCommand: ctx.runtimeCommandSpec?.installCommand,
detectCommand: ctx.runtimeCommandSpec?.detectCommand,
cwd,
env: runtimeEnv,
timeoutSec,
graceSec,
onLog,
});
await ensureAdapterExecutionTargetCommandResolvable(command, executionTarget, cwd, runtimeEnv);
const resolvedCommand = await resolveAdapterExecutionTargetCommandForLogs(command, executionTarget, cwd, runtimeEnv);
let loggedEnv = buildInvocationEnvForLogs(env, {
@@ -276,14 +300,11 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
resolvedCommand,
});
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
const extraArgs = (() => {
const fromExtraArgs = asStringArray(config.extraArgs);
if (fromExtraArgs.length > 0) return fromExtraArgs;
return asStringArray(config.args);
})();
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
let restoreRemoteWorkspace: (() => Promise<void>) | null = null;
let remoteSkillsDir: string | null = null;
let localSkillsDir: string | null = null;
@@ -531,7 +552,21 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
};
}
const clearSessionForTurnLimit = isGeminiTurnLimitResult(attempt.parsed.resultEvent, attempt.proc.exitCode);
const parsedError = typeof attempt.parsed.errorMessage === "string" ? attempt.parsed.errorMessage.trim() : "";
const stderrLine = firstNonEmptyLine(attempt.proc.stderr);
const structuredFailure = attempt.parsed.resultEvent
? describeGeminiFailure(attempt.parsed.resultEvent)
: null;
const fallbackErrorMessage =
parsedError ||
structuredFailure ||
stderrLine ||
`Gemini exited with code ${attempt.proc.exitCode ?? -1}`;
const failed = (attempt.proc.exitCode ?? 0) !== 0;
const clearSessionForTurnLimit = isGeminiTurnLimitResult(
attempt.parsed.resultEvent,
attempt.proc.exitCode,
);
// On retry, don't fall back to old session ID — the old session was stale
const canFallbackToRuntimeSession = !isRetry;
@@ -551,23 +586,24 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
: {}),
} as Record<string, unknown>)
: null;
const parsedError = typeof attempt.parsed.errorMessage === "string" ? attempt.parsed.errorMessage.trim() : "";
const stderrLine = firstNonEmptyLine(attempt.proc.stderr);
const structuredFailure = attempt.parsed.resultEvent
? describeGeminiFailure(attempt.parsed.resultEvent)
: null;
const fallbackErrorMessage =
parsedError ||
structuredFailure ||
stderrLine ||
`Gemini exited with code ${attempt.proc.exitCode ?? -1}`;
const resultJson: Record<string, unknown> = {
...(attempt.parsed.resultEvent ?? {
stdout: attempt.proc.stdout,
stderr: attempt.proc.stderr,
}),
...(failed && clearSessionForTurnLimit ? { stopReason: "max_turns_exhausted" } : {}),
};
return {
exitCode: attempt.proc.exitCode,
signal: attempt.proc.signal,
timedOut: false,
errorMessage: (attempt.proc.exitCode ?? 0) === 0 ? null : fallbackErrorMessage,
errorCode: (attempt.proc.exitCode ?? 0) !== 0 && authMeta.requiresAuth ? "gemini_auth_required" : null,
errorMessage: failed ? fallbackErrorMessage : null,
errorCode: failed && authMeta.requiresAuth
? "gemini_auth_required"
: failed && clearSessionForTurnLimit
? "max_turns_exhausted"
: null,
usage: attempt.parsed.usage,
sessionId: resolvedSessionId,
sessionParams: resolvedSessionParams,
@@ -577,10 +613,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
model,
billingType,
costUsd: attempt.parsed.costUsd,
resultJson: attempt.parsed.resultEvent ?? {
stdout: attempt.proc.stdout,
stderr: attempt.proc.stderr,
},
resultJson,
summary: attempt.parsed.summary,
question: attempt.parsed.question,
clearSession: clearSessionForTurnLimit || Boolean(clearSessionOnMissingSession && !resolvedSessionId),

View File

@@ -0,0 +1,46 @@
import { describe, expect, it } from "vitest";
import { parseGeminiJsonl } from "./parse.js";
describe("parseGeminiJsonl", () => {
it("collects assistant text from message events with string content", () => {
const stdout = [
'{"type":"init","session_id":"session-1"}',
'{"type":"message","role":"user","content":"Respond with hello."}',
'{"type":"message","role":"assistant","content":"hello","delta":true}',
'{"type":"result","status":"success"}',
].join("\n");
const parsed = parseGeminiJsonl(stdout);
expect(parsed.sessionId).toBe("session-1");
expect(parsed.summary).toBe("hello");
expect(parsed.errorMessage).toBeNull();
});
it("collects assistant text from message events with structured object content", () => {
const stdout = [
'{"type":"init","session_id":"session-2"}',
'{"type":"message","role":"assistant","content":{"content":[{"type":"text","text":"first part"},{"type":"text","text":"second part"}]}}',
'{"type":"result","status":"success"}',
].join("\n");
const parsed = parseGeminiJsonl(stdout);
expect(parsed.sessionId).toBe("session-2");
expect(parsed.summary).toBe("first part\n\nsecond part");
expect(parsed.errorMessage).toBeNull();
});
it("ignores non-assistant message events", () => {
const stdout = [
'{"type":"message","role":"user","content":"hidden user input"}',
'{"type":"message","role":"system","content":"hidden system note"}',
'{"type":"message","role":"assistant","content":"visible response"}',
'{"type":"result","status":"success"}',
].join("\n");
const parsed = parseGeminiJsonl(stdout);
expect(parsed.summary).toBe("visible response");
});
});

View File

@@ -121,6 +121,19 @@ export function parseGeminiJsonl(stdout: string) {
continue;
}
if (type === "message") {
const role = asString(event.role, "").trim().toLowerCase();
if (role === "assistant") {
// Mirror the assistant-event handling above: collect every assistant
// message including deltas. Gemini CLI emits these as discrete final
// messages (one per assistant turn), not as cumulative streaming
// tokens, so collecting all of them produces the expected concatenated
// turn-by-turn summary rather than duplicated text.
messages.push(...collectMessageText(event.content));
}
continue;
}
if (type === "result") {
resultEvent = event;
accumulateUsage(usage, event.usage ?? event.usageMetadata);
@@ -273,9 +286,18 @@ export function isGeminiTurnLimitResult(
if (exitCode === 53) return true;
if (!parsed) return false;
const status = asString(parsed.status, "").trim().toLowerCase();
if (status === "turn_limit" || status === "max_turns") return true;
const structuredStopReasons = [
parsed.status,
parsed.stopReason,
parsed.stop_reason,
parsed.errorCode,
parsed.error_code,
].map((value) => asString(value, "").trim().toLowerCase());
const error = asString(parsed.error, "").trim();
return /turn\s*limit|max(?:imum)?\s+turns?/i.test(error);
return structuredStopReasons.some((reason) =>
reason === "turn_limit" ||
reason === "max_turns" ||
reason === "max_turns_exhausted" ||
reason === "turn_limit_exhausted",
);
}

View File

@@ -55,8 +55,6 @@ export function buildGeminiLocalConfig(v: CreateConfigValues): Record<string, un
const ac: Record<string, unknown> = {};
if (v.cwd) ac.cwd = v.cwd;
if (v.instructionsFilePath) ac.instructionsFilePath = v.instructionsFilePath;
if (v.promptTemplate) ac.promptTemplate = v.promptTemplate;
if (v.bootstrapPrompt) ac.bootstrapPromptTemplate = v.bootstrapPrompt;
ac.model = v.model || DEFAULT_GEMINI_LOCAL_MODEL;
ac.timeoutSec = 0;
ac.graceSec = 15;

View File

@@ -1,8 +1,17 @@
import type { AdapterModelProfileDefinition } from "@paperclipai/adapter-utils";
export const type = "opencode_local";
export const label = "OpenCode (local)";
export const DEFAULT_OPENCODE_LOCAL_MODEL = "openai/gpt-5.2-codex";
export function isValidOpenCodeModelId(value: unknown): value is string {
if (typeof value !== "string") return false;
const trimmed = value.trim();
const slashIndex = trimmed.indexOf("/");
return Boolean(trimmed) && slashIndex > 0 && slashIndex !== trimmed.length - 1;
}
export const models: Array<{ id: string; label: string }> = [
{ id: DEFAULT_OPENCODE_LOCAL_MODEL, label: DEFAULT_OPENCODE_LOCAL_MODEL },
{ id: "openai/gpt-5.4", label: "openai/gpt-5.4" },
@@ -11,6 +20,19 @@ export const models: Array<{ id: string; label: string }> = [
{ id: "openai/gpt-5.1-codex-mini", label: "openai/gpt-5.1-codex-mini" },
];
export const modelProfiles: AdapterModelProfileDefinition[] = [
{
key: "cheap",
label: "Cheap",
description: "Use OpenCode's known Codex mini model as the budget lane.",
adapterConfig: {
model: "openai/gpt-5.1-codex-mini",
variant: "low",
},
source: "adapter_default",
},
];
export const agentConfigurationDoc = `# opencode_local agent configuration
Adapter: opencode_local

View File

@@ -11,24 +11,38 @@ const {
restoreWorkspaceFromSshExecution,
runSshCommand,
syncDirectoryToSsh,
startAdapterExecutionTargetPaperclipBridge,
} = vi.hoisted(() => ({
runChildProcess: vi.fn(async () => ({
exitCode: 0,
signal: null,
timedOut: false,
stdout: [
JSON.stringify({ type: "step_start", sessionID: "session_123" }),
JSON.stringify({ type: "text", sessionID: "session_123", part: { text: "hello" } }),
JSON.stringify({
type: "step_finish",
sessionID: "session_123",
part: { cost: 0.001, tokens: { input: 1, output: 1, reasoning: 0, cache: { read: 0, write: 0 } } },
}),
].join("\n"),
stderr: "",
pid: 123,
startedAt: new Date().toISOString(),
})),
runChildProcess: vi.fn(async (_runId: string, _command: string, args: string[]) => {
if (args.includes("models")) {
return {
exitCode: 0,
signal: null,
timedOut: false,
stdout: "opencode/gpt-5-nano\nopenai/gpt-4.1\n",
stderr: "",
pid: 122,
startedAt: new Date().toISOString(),
};
}
return {
exitCode: 0,
signal: null,
timedOut: false,
stdout: [
JSON.stringify({ type: "step_start", sessionID: "session_123" }),
JSON.stringify({ type: "text", sessionID: "session_123", part: { text: "hello" } }),
JSON.stringify({
type: "step_finish",
sessionID: "session_123",
part: { cost: 0.001, tokens: { input: 1, output: 1, reasoning: 0, cache: { read: 0, write: 0 } } },
}),
].join("\n"),
stderr: "",
pid: 123,
startedAt: new Date().toISOString(),
};
}),
ensureCommandResolvable: vi.fn(async () => undefined),
resolveCommandForLogs: vi.fn(async () => "ssh://fixture@127.0.0.1:2222/remote/workspace :: opencode"),
prepareWorkspaceForSshExecution: vi.fn(async () => undefined),
@@ -39,6 +53,14 @@ const {
exitCode: 0,
})),
syncDirectoryToSsh: vi.fn(async () => undefined),
startAdapterExecutionTargetPaperclipBridge: vi.fn(async () => ({
env: {
PAPERCLIP_API_URL: "http://127.0.0.1:4310",
PAPERCLIP_API_KEY: "bridge-token",
PAPERCLIP_API_BRIDGE_MODE: "queue_v1",
},
stop: async () => {},
})),
}));
vi.mock("@paperclipai/adapter-utils/server-utils", async () => {
@@ -66,6 +88,16 @@ vi.mock("@paperclipai/adapter-utils/ssh", async () => {
};
});
vi.mock("@paperclipai/adapter-utils/execution-target", async () => {
const actual = await vi.importActual<typeof import("@paperclipai/adapter-utils/execution-target")>(
"@paperclipai/adapter-utils/execution-target",
);
return {
...actual,
startAdapterExecutionTargetPaperclipBridge,
};
});
import { execute } from "./execute.js";
describe("opencode remote execution", () => {
@@ -84,7 +116,9 @@ describe("opencode remote execution", () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-opencode-remote-"));
cleanupDirs.push(rootDir);
const workspaceDir = path.join(rootDir, "workspace");
const alternateWorkspaceDir = path.join(rootDir, "workspace-other");
await mkdir(workspaceDir, { recursive: true });
await mkdir(alternateWorkspaceDir, { recursive: true });
const result = await execute({
runId: "run-1",
@@ -110,6 +144,20 @@ describe("opencode remote execution", () => {
cwd: workspaceDir,
source: "project_primary",
},
paperclipWorkspaces: [
{
workspaceId: "workspace-1",
cwd: workspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
cwd: alternateWorkspaceDir,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
],
},
executionTransport: {
remoteExecution: {
@@ -121,7 +169,6 @@ describe("opencode remote execution", () => {
privateKey: "PRIVATE KEY",
knownHosts: "[127.0.0.1]:2222 ssh-ed25519 AAAA",
strictHostKeyChecking: true,
paperclipApiUrl: "http://198.51.100.10:3102",
},
},
onLog: async () => {},
@@ -136,7 +183,6 @@ describe("opencode remote execution", () => {
port: 2222,
username: "fixture",
remoteCwd: "/remote/workspace",
paperclipApiUrl: "http://198.51.100.10:3102",
},
});
expect(prepareWorkspaceForSshExecution).toHaveBeenCalledTimes(1);
@@ -153,15 +199,105 @@ describe("opencode remote execution", () => {
expect.stringContaining(".claude/skills"),
expect.anything(),
);
const call = runChildProcess.mock.calls[0] as unknown as
const runCall = runChildProcess.mock.calls.find((entry) => Array.isArray(entry[2]) && entry[2].includes("run")) as
| [string, string, string[], { env: Record<string, string>; remoteExecution?: { remoteCwd: string } | null }]
| undefined;
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://198.51.100.10:3102");
const modelProbeCall = runChildProcess.mock.calls.find((entry) => Array.isArray(entry[2]) && entry[2].includes("models")) as
| [string, string, string[], { env: Record<string, string>; remoteExecution?: { remoteCwd: string } | null }]
| undefined;
expect(modelProbeCall?.[2]).toEqual(["models"]);
expect(modelProbeCall?.[3].env.XDG_CONFIG_HOME).toBe(
"/remote/workspace/.paperclip-runtime/opencode/xdgConfig",
);
expect(modelProbeCall?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
const call = runCall as
| [string, string, string[], { env: Record<string, string>; remoteExecution?: { remoteCwd: string } | null }]
| undefined;
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe("/remote/workspace");
expect(JSON.parse(call?.[3].env.PAPERCLIP_WORKSPACES_JSON ?? "[]")).toEqual([
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
{
workspaceId: "workspace-2",
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "feature/other",
},
]);
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:4310");
expect(call?.[3].env.PAPERCLIP_API_BRIDGE_MODE).toBe("queue_v1");
expect(call?.[3].env.XDG_CONFIG_HOME).toBe("/remote/workspace/.paperclip-runtime/opencode/xdgConfig");
expect(call?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
expect(startAdapterExecutionTargetPaperclipBridge).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledTimes(1);
});
it("fails before the remote run when the configured model is unavailable on the SSH target", async () => {
runChildProcess.mockImplementationOnce(async () => ({
exitCode: 0,
signal: null,
timedOut: false,
stdout: "openai/gpt-4.1\n",
stderr: "",
pid: 456,
startedAt: new Date().toISOString(),
}));
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-opencode-remote-model-"));
cleanupDirs.push(rootDir);
const workspaceDir = path.join(rootDir, "workspace");
await mkdir(workspaceDir, { recursive: true });
await expect(() =>
execute({
runId: "run-ssh-model-missing",
agent: {
id: "agent-1",
companyId: "company-1",
name: "OpenCode Builder",
adapterType: "opencode_local",
adapterConfig: {},
},
runtime: {
sessionId: null,
sessionParams: null,
sessionDisplayId: null,
taskKey: null,
},
config: {
command: "opencode",
model: "opencode/gpt-5-nano",
},
context: {
paperclipWorkspace: {
cwd: workspaceDir,
source: "project_primary",
},
},
executionTransport: {
remoteExecution: {
host: "127.0.0.1",
port: 2222,
username: "fixture",
remoteWorkspacePath: "/remote/workspace",
remoteCwd: "/remote/workspace",
privateKey: "PRIVATE KEY",
knownHosts: "[127.0.0.1]:2222 ssh-ed25519 AAAA",
strictHostKeyChecking: true,
},
},
onLog: async () => {},
}),
).rejects.toThrow("Configured OpenCode model is unavailable on the remote execution target");
expect(runChildProcess).toHaveBeenCalledTimes(1);
expect((runChildProcess.mock.calls[0]?.[2] as string[] | undefined) ?? []).toEqual(["models"]);
expect(startAdapterExecutionTargetPaperclipBridge).not.toHaveBeenCalled();
});
it("resumes saved OpenCode sessions for remote SSH execution only when the identity matches", async () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-opencode-remote-resume-"));
cleanupDirs.push(rootDir);
@@ -218,7 +354,9 @@ describe("opencode remote execution", () => {
onLog: async () => {},
});
const call = runChildProcess.mock.calls[0] as unknown as [string, string, string[]] | undefined;
const call = runChildProcess.mock.calls.find((entry) => Array.isArray(entry[2]) && entry[2].includes("run")) as
| [string, string, string[]]
| undefined;
expect(call?.[2]).toContain("--session");
expect(call?.[2]).toContain("session-123");
});

View File

@@ -5,7 +5,6 @@ import { fileURLToPath } from "node:url";
import { inferOpenAiCompatibleBiller, type AdapterExecutionContext, type AdapterExecutionResult } from "@paperclipai/adapter-utils";
import {
adapterExecutionTargetIsRemote,
adapterExecutionTargetPaperclipApiUrl,
adapterExecutionTargetRemoteCwd,
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetSessionMatches,
@@ -13,6 +12,7 @@ import {
adapterExecutionTargetUsesPaperclipBridge,
describeAdapterExecutionTarget,
ensureAdapterExecutionTargetCommandResolvable,
ensureAdapterExecutionTargetRuntimeCommandInstalled,
prepareAdapterExecutionTargetRuntime,
readAdapterExecutionTarget,
readAdapterExecutionTargetHomeDir,
@@ -35,6 +35,7 @@ import {
ensurePathInEnv,
renderTemplate,
renderPaperclipWakePrompt,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
runChildProcess,
@@ -42,7 +43,11 @@ import {
resolvePaperclipDesiredSkillNames,
} from "@paperclipai/adapter-utils/server-utils";
import { isOpenCodeUnknownSessionError, parseOpenCodeJsonl } from "./parse.js";
import { ensureOpenCodeModelConfiguredAndAvailable } from "./models.js";
import {
ensureOpenCodeModelConfiguredAndAvailable,
parseOpenCodeModelsOutput,
requireOpenCodeModelId,
} from "./models.js";
import { removeMaintainerOnlySkillSymlinks } from "@paperclipai/adapter-utils/server-utils";
import { prepareOpenCodeRuntimeConfig } from "./runtime-config.js";
@@ -68,6 +73,64 @@ function resolveOpenCodeBiller(env: Record<string, string>, provider: string | n
return inferOpenAiCompatibleBiller(env, null) ?? provider ?? "unknown";
}
const REMOTE_OPENCODE_MODELS_PROBE_DEFAULT_TIMEOUT_SEC = 20;
async function ensureRemoteOpenCodeModelConfiguredAndAvailable(input: {
runId: string;
executionTarget: NonNullable<AdapterExecutionContext["executionTarget"]>;
command: string;
model: string;
cwd: string;
env: Record<string, string>;
timeoutSec: number;
graceSec: number;
}) {
const model = requireOpenCodeModelId(input.model);
const probeTimeoutSec = input.timeoutSec > 0
? Math.min(input.timeoutSec, REMOTE_OPENCODE_MODELS_PROBE_DEFAULT_TIMEOUT_SEC)
: REMOTE_OPENCODE_MODELS_PROBE_DEFAULT_TIMEOUT_SEC;
const probe = await runAdapterExecutionTargetProcess(
input.runId,
input.executionTarget,
input.command,
["models"],
{
cwd: input.cwd,
env: input.env,
timeoutSec: probeTimeoutSec,
graceSec: input.graceSec,
onLog: async () => {},
},
);
if (probe.timedOut) {
throw new Error(`\`opencode models\` timed out on the remote execution target after ${probeTimeoutSec}s.`);
}
if ((probe.exitCode ?? 1) !== 0) {
const detail = firstNonEmptyLine(probe.stderr) || firstNonEmptyLine(probe.stdout);
throw new Error(
detail
? `\`opencode models\` failed on the remote execution target: ${detail}`
: "`opencode models` failed on the remote execution target.",
);
}
const models = parseOpenCodeModelsOutput(probe.stdout);
if (models.length === 0) {
throw new Error(
"OpenCode returned no models on the remote execution target. Run `opencode models` there and verify provider auth.",
);
}
if (!models.some((entry) => entry.id === model)) {
const sample = models.slice(0, 12).map((entry) => entry.id).join(", ");
throw new Error(
`Configured OpenCode model is unavailable on the remote execution target: ${model}. Available models: ${sample}${models.length > 12 ? ", ..." : ""}`,
);
}
}
function claudeSkillsHome(): string {
return path.join(os.homedir(), ".claude", "skills");
}
@@ -155,6 +218,13 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
const useConfiguredInsteadOfAgentHome = workspaceSource === "agent_home" && configuredCwd.length > 0;
const effectiveWorkspaceCwd = useConfiguredInsteadOfAgentHome ? "" : workspaceCwd;
const cwd = effectiveWorkspaceCwd || configuredCwd || process.cwd();
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const shapedWorkspaceEnv = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: effectiveWorkspaceCwd,
workspaceHints,
executionTargetIsRemote,
executionCwd: effectiveExecutionCwd,
});
await ensureAbsoluteDirectory(cwd, { createIfMissing: true });
const openCodeSkillEntries = await readPaperclipRuntimeSkillEntries(config, __moduleDir);
const desiredOpenCodeSkillNames = resolvePaperclipDesiredSkillNames(config, openCodeSkillEntries);
@@ -203,17 +273,16 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
if (linkedIssueIds.length > 0) env.PAPERCLIP_LINKED_ISSUE_IDS = linkedIssueIds.join(",");
if (wakePayloadJson) env.PAPERCLIP_WAKE_PAYLOAD_JSON = wakePayloadJson;
applyPaperclipWorkspaceEnv(env, {
workspaceCwd: effectiveWorkspaceCwd,
workspaceCwd: shapedWorkspaceEnv.workspaceCwd,
workspaceSource,
workspaceId,
workspaceRepoUrl,
workspaceRepoRef,
agentHome,
});
if (workspaceHints.length > 0) env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(workspaceHints);
const targetPaperclipApiUrl = adapterExecutionTargetPaperclipApiUrl(executionTarget);
if (targetPaperclipApiUrl) env.PAPERCLIP_API_URL = targetPaperclipApiUrl;
if (shapedWorkspaceEnv.workspaceHints.length > 0) {
env.PAPERCLIP_WORKSPACES_JSON = JSON.stringify(shapedWorkspaceEnv.workspaceHints);
}
for (const [key, value] of Object.entries(envConfig)) {
if (typeof value === "string") env[key] = value;
}
@@ -234,6 +303,19 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
(entry): entry is [string, string] => typeof entry[1] === "string",
),
);
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
await ensureAdapterExecutionTargetRuntimeCommandInstalled({
runId,
target: executionTarget,
installCommand: ctx.runtimeCommandSpec?.installCommand,
detectCommand: ctx.runtimeCommandSpec?.detectCommand,
cwd,
env: runtimeEnv,
timeoutSec,
graceSec,
onLog,
});
await ensureAdapterExecutionTargetCommandResolvable(command, executionTarget, cwd, runtimeEnv);
const resolvedCommand = await resolveAdapterExecutionTargetCommandForLogs(command, executionTarget, cwd, runtimeEnv);
let loggedEnv = buildInvocationEnvForLogs(preparedRuntimeConfig.env, {
@@ -241,7 +323,6 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
includeRuntimeKeys: ["HOME"],
resolvedCommand,
});
if (!executionTargetIsRemote) {
await ensureOpenCodeModelConfiguredAndAvailable({
model,
@@ -251,20 +332,17 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
});
}
const timeoutSec = asNumber(config.timeoutSec, 0);
const graceSec = asNumber(config.graceSec, 20);
const extraArgs = (() => {
const fromExtraArgs = asStringArray(config.extraArgs);
if (fromExtraArgs.length > 0) return fromExtraArgs;
return asStringArray(config.args);
})();
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
let restoreRemoteWorkspace: (() => Promise<void>) | null = null;
let localSkillsDir: string | null = null;
let remoteRuntimeRootDir: string | null = null;
let paperclipBridge: Awaited<ReturnType<typeof startAdapterExecutionTargetPaperclipBridge>> = null;
if (executionTargetIsRemote) {
if (executionTarget?.kind === "remote") {
localSkillsDir = await buildOpenCodeSkillsDir(config);
await onLog(
"stdout",
@@ -315,6 +393,16 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
{ cwd, env: preparedRuntimeConfig.env, timeoutSec, graceSec, onLog },
);
}
await ensureRemoteOpenCodeModelConfiguredAndAvailable({
runId,
executionTarget,
command,
model,
cwd,
env: preparedRuntimeConfig.env,
timeoutSec,
graceSec,
});
}
if (executionTargetIsRemote && adapterExecutionTargetUsesPaperclipBridge(executionTarget)) {
paperclipBridge = await startAdapterExecutionTargetPaperclipBridge({

View File

@@ -67,6 +67,7 @@ export {
listOpenCodeModels,
discoverOpenCodeModels,
ensureOpenCodeModelConfiguredAndAvailable,
requireOpenCodeModelId,
resetOpenCodeModelsCacheForTests,
} from "./models.js";
export { parseOpenCodeJsonl, isOpenCodeUnknownSessionError } from "./parse.js";

Some files were not shown because too many files have changed in this diff Show More