Add @deprecated JSDoc and inline comments to bootstrapPromptTemplate
references in agent-instructions and company-portability services.
This field is superseded by the managed instructions bundle system.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- SPEC: reflect that Paperclip now manages task-linked documents and
attachments (issue documents, file attachments) instead of claiming
it does not manage work artifacts
- agents-runtime: remove bootstrapPromptTemplate from recommended config,
add deprecation notice, update minimal setup checklist
Co-Authored-By: Paperclip <noreply@paperclip.ing>
New CEO agents created during onboarding now include explicit delegation
rules: triage tasks, route to CTO/CMO/UXDesigner, never do IC work, and
follow up on delegated work.
* ci: add Dockerfile deps stage validation to PR policy
Checks that all workspace package.json files and the patches/
directory are copied into the Dockerfile deps stage. Prevents the
Docker build from breaking when new packages or patches are added
without updating the Dockerfile.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* ci: scope Dockerfile check to deps stage and derive workspace roots
Address Greptile review feedback:
- Use awk to extract only the deps stage before grepping, preventing
false positives from COPY lines in other stages
- Derive workspace search roots from pnpm-workspace.yaml instead of
hardcoding them, so new top-level workspaces are automatically covered
* ci: guard against empty workspace roots in Dockerfile check
Fail early if pnpm-workspace.yaml parsing yields no search roots,
preventing a silent false-pass from find defaulting to cwd.
* ci: guard against empty deps stage extraction
Fail early with a clear error if awk cannot find the deps stage in the
Dockerfile, instead of producing misleading "missing COPY" errors.
* ci: deduplicate find results from overlapping workspace roots
Use sort -u instead of sort to prevent duplicate error messages when
nested workspace globs (e.g. packages/* and packages/adapters/*) cause
the same package.json to be found twice.
* ci: anchor grep to ^COPY to ignore commented-out Dockerfile lines
Prevents false negatives when a COPY directive is commented out
(e.g. # COPY packages/foo/package.json).
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
When CURSOR_API_KEY is not set, check ~/.cursor/cli-config.json for
authInfo from `agent login` before emitting the missing key warning.
Users authenticated via native login no longer see a false warning.
The previous documentation parenthetical "(defaulting to ~/.codex/skills/)"
was misleading because Paperclip almost always sets CODEX_HOME to a
per-company managed home. Update index.ts docs, skills.ts detail string,
and execute.ts inline comment to make the runtime path unambiguous.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Setting the env var before the user-config loop meant adapter env
overrides could disable the guard. Move it after the loop so it
always wins, matching the pattern already used in test.ts and
models.ts.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Pi outputs the model list to stderr instead of stdout. This fix checks
stderr first and falls back to stdout for compatibility with older
versions.
Fixes model discovery returning empty arrays and environment tests
failing with 'Pi returned no models' error.
Set OPENCODE_DISABLE_PROJECT_CONFIG=true in all OpenCode invocations
(execute, model discovery, environment test) to stop the OpenCode CLI
from writing an opencode.json file into the project working directory.
Model selection is already passed via the --model CLI flag.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The default fallback in ensureCodexSkillsInjected still referenced the
old function name. Updated to use resolveCodexSkillsDir with shared
home as fallback.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The previous commit incorrectly used resolveSharedCodexHomeDir() (~/.codex)
but Codex runs with CODEX_HOME set to a per-company managed home under
~/.paperclip/instances/. Skills injected into ~/.codex/skills/ would not
be discoverable by Codex. Now uses effectiveCodexHome directly.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The Codex adapter was the only one injecting skills into
<cwd>/.agents/skills/, polluting the project's git repo. All other
adapters (Gemini, Cursor, etc.) use a home-based directory. This
changes the Codex adapter to inject into ~/.codex/skills/ (resolved
via resolveSharedCodexHomeDir) to match the established pattern.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
pnpm install needs the patches/ directory to resolve patched
dependencies (embedded-postgres). Without it, --frozen-lockfile
fails with ENOENT on the patch file.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use path.join instead of string concatenation for the auth.json
fallback path in the detail message, ensuring correct path
separators on Windows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When adapter config has no CODEX_HOME but process.env.CODEX_HOME is
set, readCodexAuthInfo reads from the process env path. The detail
message now uses codexHomeDir() instead of hardcoded "~/.codex" so
the displayed path always matches where credentials were read from.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Consolidate the env stub into beforeEach so the pre-existing cwd
test is also isolated from host OPENAI_API_KEY, avoiding
non-deterministic filesystem side effects.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Show the configured CODEX_HOME path instead of hardcoded ~/.codex
when the email fallback message is displayed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use vi.stubEnv to clear OPENAI_API_KEY in both new tests so they
don't silently pass the wrong branch when the key is set in the
test runner's environment.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rebuild the patch file with valid unified-diff hunks so pnpm can
apply the locale and environment fixes during install.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The vendor package still hardcoded LC_MESSAGES to en_US.UTF-8.
That locale is missing in slim containers, and initdb fails during
bootstrap even when --lc-messages=C is passed later.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Use globalThis.process.env in the vendor patch so the spawned child
process config does not trip over the local process binding inside
embedded-postgres.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Add tests for codex_native_auth_present and codex_openai_api_key_missing
code paths. Also pass adapter-configured CODEX_HOME through to
readCodexAuthInfo so the probe respects per-adapter home directories.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Change vertical-align from baseline to middle on both editor and
read-only mention chip styles. The baseline alignment caused
inconsistent positioning because the agent ::before icon (0.75rem)
and project ::before dot (0.45rem) produced different synthesized
baselines in the inline-flex containers.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The embedded-postgres library hardcodes --lc-messages=en_US.UTF-8 and
strips the parent process environment when spawning initdb/postgres.
In slim Docker images (e.g. node:20-bookworm-slim), the en_US.UTF-8
locale isn't installed, causing initdb to exit with code 1.
Two fixes applied:
1. Add --lc-messages=C to all initdbFlags arrays (overrides the
library's hardcoded locale since our flags come after in the spread)
2. pnpm patch on embedded-postgres to preserve process.env in spawn
calls, preventing loss of PATH, LD_LIBRARY_PATH, and other vars
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Join requests were displayed in a separate card-style section below the main
inbox list. This moves them into the unified work items feed so they sort
chronologically alongside issues, approvals, and failed runs—matching the
inline treatment hiring requests already receive.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Loading an instructions file is normal, expected behavior — not worth
logging to stdout/stderr on every run. Warning logs for failed reads
are preserved.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The environment test warned about OPENAI_API_KEY being unset even
when Codex was authenticated via `codex auth`. Now checks
~/.codex/auth.json before emitting the warning.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Drop entities package (CI blocks pnpm-lock.yaml on PRs; reset lockfile to master)
- Restore numeric + allowlisted named entity decoding in issues.ts
- Split Greptile mid-token & case into its own test with review comment
Made-with: Cursor
Addresses Greptile review on PR #1363: numeric entities decode via
code points; named entities use a small allowlist (amp, nbsp, etc.)
so M&M resolves correctly; unknown named entities are preserved.
Adds mid-token tests for & in agent names.
Made-with: Cursor
The plugin framework landed without updating the Dockerfile. The
server now imports @paperclipai/plugin-sdk, so the deps stage needs
its package.json for install and the build stage needs to compile
it before building the server.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Address Greptile feedback — the sparse 3-line placeholder could
lead to shallow thinking paths. Expanded to 6 lines with guiding
brackets and added "Aim for 5–8 steps" hint in the comment.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
PAPERCLIP_MIGRATION_PROMPT=never was checked before
PAPERCLIP_MIGRATION_AUTO_APPLY=true, causing auto-apply to never
trigger when both env vars are set (as in dev:watch). Swap the
check order so AUTO_APPLY takes precedence.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Prevent company imports from re-enabling scheduler heartbeats on imported agents and cover both new-company and existing-company import flows in portability tests.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Documents the `paperclipai company export` and `paperclipai company import`
CLI commands, covering package format, all options, target modes, collision
strategies, GitHub sources, interactive selection, and API endpoints.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Infer portable repo metadata from local git workspaces when repoUrl is missing, and collapse repeated task workspace export warnings into a single summary per missing workspace.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Adds a destructive-variant button at the top of the heartbeats page that
disables timer heartbeats for all agents at once. The button only appears
when at least one agent has heartbeats enabled, and shows a loading state
while processing.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Agents without a heartbeat interval configured (intervalSec=0) were
filtered out, making them invisible on the instance heartbeats page.
This prevented managing heartbeats for agents that hadn't been
configured yet (e.g. donchitos company agents).
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The /instance/scheduler-heartbeats endpoint was filtering agents by the
requesting user's company memberships, which meant non-member companies
(like donchitos) were hidden. Since this is an instance-level admin page,
it should show all agents across all companies.
- Added assertInstanceAdmin to authz.ts for reuse
- Replaced assertBoard + company filter with assertInstanceAdmin
- Removed the companyIds-based WHERE clause since instance admins see all
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Persist realized isolated/operator workspaces back onto the issue as reusable workspaces so later runs stay on the same workspace, and update the issue workspace picker to present realized isolated workspaces as existing workspaces.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Pills with semi-transparent backgrounds were using raw color luminance to pick
text color, ignoring the page background showing through. This caused unreadable
text on dark themes for mid-luminance colors like orange. Now composites the
rgba background over the actual page bg (dark/light) before computing WCAG
contrast ratios, and centralizes the logic in a shared color-contrast utility.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Shorten button labels ("New document" → "New", "Upload attachment" → "Upload") on small screens
- Add min-w-0 and shrink-0 to flex containers and items to prevent overflow
- Truncate document revision text on narrow viewports
- Mark chevron, key badge, and action buttons as shrink-0
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Sort the flat list view, org tree view, and sidebar agents list
alphabetically by name using localeCompare.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The initial mobile toggle fix (afc3d7ec) missed 2 toggle switches in
RoutineDetail.tsx and Routines.tsx. Without the attribute, these toggles
would still inflate to 44px on touch devices via the @media (pointer: coarse)
rule.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The global @media (pointer: coarse) rule was forcing min-height: 44px on
toggle button elements, inflating them from 20px to 44px. Added
data-slot="toggle" to all 10 toggle buttons and a CSS override to reset
their min-height, keeping the parent row as the touch target.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
On mobile, the two-panel file browser + editor layout now stacks
vertically with a toggleable file panel. The draggable separator is
hidden, and selecting a file auto-closes the panel to maximize
editor space.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The previous fix (validateUrl on linkPlugin) only affected the link dialog,
not the markdown-to-Lexical import path. Lexical's LinkNode.sanitizeUrl()
converts agent:// and project:// URLs to about:blank because they aren't
in its allowlist. Override the prototype method to preserve these schemes
so mention chips render correctly.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Add name, id, and htmlFor attributes to form inputs and a method/action
to the form element so password managers can properly identify the login
form fields.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
1.1em was too large per feedback — settle on 1em for all markdown
monospace contexts.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Bump monospace font-size from 0.78em to 1.1em across all markdown
contexts (editor, code blocks, inline code). Add subtle gray
background (#ffffff0f) for inline code in dark mode.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Priority is still supported as a feature (editable in issue properties,
used in filters), but no longer shown prominently in every issue row.
Affects inbox, issues list, my issues, and dashboard pages.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
**#1 — Missing `description` field in fields table**
The create body example included `description` and the schema confirms
`description: z.string().optional().nullable()`, but the reference table
omitted it. Added as an optional field.
**#2 — Concurrency policy descriptions were inaccurate**
Original docs described both `coalesce_if_active` and `skip_if_active` as
variants of "skip", which was wrong. Source-verified against
`server/src/services/routines.ts` (dispatchRoutineRun, line 568):
const status = concurrencyPolicy === "skip_if_active" ? "skipped" : "coalesced";
Both policies write identical DB state (same linkedIssueId and
coalescedIntoRunId); the only difference is the run status value.
Descriptions now reflect this: both finalise the incoming run immediately
and link it to the active run — no new issue is created in either case.
Note: the reviewer's suggestion that `coalesce_if_active` "extends or
notifies" the active run was also not supported by the code; corrected
accordingly.
**#3 — `triggerId` undocumented in Manual Run**
`runRoutineSchema` accepts `triggerId` and the service genuinely uses it
(routines.ts:1029–1034): fetches the trigger, enforces that it belongs to
the routine (403) and is enabled (409), then passes it to dispatchRoutineRun
which records the run against the trigger and updates its `lastFiredAt`.
Added `triggerId` to the example body and documented all three behaviours.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Routines are recurring tasks that fire on a schedule, webhook, or API
call and create a heartbeat run for the assigned agent. Document the
full CRUD surface including:
- List / get routines
- Create with concurrency and catch-up policy options
- Add schedule, webhook, and api triggers
- Update / delete triggers, rotate webhook secrets
- Manual run and public trigger fire
- List run history
- Agent access rules (agents can only manage own routines)
- Routine lifecycle (active → paused → archived)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Fix goals-and-projects.md: `completed` is not a valid status — correct to
`achieved` and document all valid values (planned/active/achieved/cancelled)
- Fix issues.md: document that `expectedStatuses: ["in_progress"]` can be used
to re-claim a stale lock after a crashed run; clarify that `runId` in the
request body is not accepted (run ID comes from X-Paperclip-Run-Id header only)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Merge pr-verify.yml, pr-policy.yml, and pr-e2e.yml into a single
pr.yml with three parallel jobs (policy, verify, e2e). Benefits:
- Single concurrency group cancels all jobs on new push
- Consistent Node 24 across all jobs
- One file to maintain instead of three
The jobs still run independently (no artifact sharing) since pnpm
cache makes install fast and the upload/download overhead for
node_modules would negate the savings.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Align with e2e.yml and ensure CI tests exactly the committed
dependency tree. The pr-policy job already blocks lockfile changes
in PRs, so frozen-lockfile is safe here.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The playwright.config.ts had `reuseExistingServer: !!process.env.CI`
which meant CI would reuse (expect) an existing server while local
dev would start one. This is backwards — in CI Playwright should
manage the server, and in local dev you likely already have one
running.
Flip to `!process.env.CI` and remove the `CI: ""` env override
from the workflow.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Enhanced button and text elements to ensure proper overflow handling and truncation for agent names and statuses.
- Adjusted class names for better responsiveness and visual consistency, particularly for unknown and terminated managers.
- Added handling for cases where the selected manager is terminated, displaying a distinct style and message.
- Introduced a new state for unknown managers, providing user feedback when the saved manager is missing.
- Improved layout for displaying current manager status, ensuring clarity in the UI.
- Modified the display of the current agent's name to include a "(terminated)" suffix if the agent's status is terminated.
- Adjusted button layout to ensure proper text truncation and overflow handling for agent names and roles.
- Introduced ReportsToPicker component in AgentConfigForm and NewAgent pages to allow selection of an agent's manager.
- Updated organizational structure documentation to reflect the ability to change an agent's manager post-creation.
- Enhanced error handling in ConfigurationTab to provide user feedback on save failures.
mkdir -p the CODEX_HOME directory in codex-local adapter and the
agentHome directory in the heartbeat service before passing them to
adapters. This prevents CLI tools from erroring when their home
directory hasn't been created yet. Covers all local adapters that
set AGENT_HOME.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Adds an amber "Beta" tag next to "Routines" in both the page heading
and the sidebar navigation item. Extends SidebarNavItem with textBadge
and textBadgeTone props for reusable text badges.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rich-text comments store entities like   after @names; strip them before matching agents so issue_comment_mentioned and wake injection work.
Made-with: Cursor
Issues created from routine executions now display a clickable "Routine"
badge in the header bar, linking back to the originating routine.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Replace the button ("Enabled"/"Disabled") for "Can create new agents" and
the custom oversized switch for "Can assign tasks" with the same toggle
style (h-5 w-9, bg-green-600/bg-muted) used throughout Run Policy.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The access service mock was missing the setPrincipalPermission function,
causing 5 test failures in the import flow.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
GitHub recommends 1280x640 for repository social media previews.
The org chart SVG/PNG now always outputs at these dimensions,
scaling and centering the content to fit any org size.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Share issue-assignment wakeup logic between direct issue creation and routine-created execution issues, and add regression coverage for the routine path.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
COMPANY.md now contains only the company description in frontmatter,
without the agents list and projects list that were previously rendered
in the markdown body. The README.md already contains this info.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Remove routine status from the editable detail draft so the active/paused switch remains an immediate save path instead of surfacing unsaved form state.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When navigating to the company export page without a specific file in
the URL, select README.md by default instead of whichever file happens
to be first in the export result (previously COMPANY.md).
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Avatar circle moved from y+24 to y+27 across all three card renderers
for balanced whitespace between card top and text baseline
- Paperclip logo icon scaled to 0.72x with adjusted text position
to better match the wordmark size
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The previous text centering fix (y+66/y+82) only updated the circuit
and schematic custom renderers. The defaultRenderCard used by warmth,
monochrome, and nebula still had the old y+52/y+68 positions.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Moved name baseline from y+58 to y+66 and role from y+74 to y+82,
centering the text block in the 55px gap between the avatar circle
bottom (y+41) and the card bottom (y+96).
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Increase card height from 88 to 96px for better text spacing
- Move name/role text down (y+58/y+74) so text sits properly below emoji
- Fix Paperclip logo watermark — vertically center text with icon
- Render PNG at 2x density (144 DPI) for retina-quality output
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Drop "api" from the trigger kind dropdown and disable the "webhook"
option with a "COMING SOON" label until it's ready.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Embeds Twemoji SVG paths directly into the org chart cards, replacing
monochrome icon paths with full-color emoji graphics (crown, laptop,
globe, keyboard, microscope, wand, chart, gear). Works in pure SVG
without any browser, emoji font, or Satori dependency.
Twemoji graphics are CC-BY 4.0 licensed (Twitter/X).
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Replace the cluttered rounded-full bordered pills around each activity
detail with clean inline text separated by dot separators. Wrap in a
bordered card container matching the runs tab style.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The editable title now sits in the header alongside Run Now and
the active/paused toggle, replacing the old "Automation" subheading.
This removes the redundant label and gives the title top-level
prominence in the routine detail view.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rewrote org-chart-svg.ts to support all 5 styles (monochrome, nebula,
circuit, warmth, schematic) as pure SVG — no browser or Satori needed.
Routes now accept ?style= query param. Added standalone comparison script.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Previously used routine run status ("received"/"issue_created") which
are not the right signal. Now queries heartbeatsApi.liveRunsForIssue()
on the active issue to check if an agent is actually running — the same
source of truth the LiveRunWidget uses.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The blue dot and LiveRunWidget were driven by `routine.activeIssue`,
which returns any open execution issue — even after the heartbeat run
finishes. Now checks routineRuns for status "received" or "issue_created"
to determine if a run is actually in progress.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Persist child-process metadata for local adapter runs, keep detached runs alive when their pid still exists, queue a single automatic retry when the pid is confirmed dead, and clear detached warnings when the original run reports activity again.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Moves the cost summary out of a collapsible section below the tabs and
into the Activity tab content, displayed as a static card at the top.
Removes the now-unused `cost` state from `secondaryOpen`.
Closes PAP-559
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Strip the rounded-border card wrappers from repo URL and local folder
fields so they sit directly in the section. Add pt-3 to the section
so the first field doesn't touch the border-t above it.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Failed runs are no longer shown in a separate section. They are now
mixed into the main work items feed sorted by timestamp, matching
how approvals are already interleaved with issues.
Replaced the large FailedRunCard with a compact FailedRunInboxRow
that matches the ApprovalInboxRow visual style.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Remove the workspace setup toggle menu ("Where will work be done on this
project?") and instead always display both repo URL and local folder
inputs directly. Both fields are marked as optional with help tooltips
explaining their purpose. Repo is shown first, local folder second.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Instead of the generic "Hire Agent" label, display the agent's name from
the approval payload for hire_agent approvals across inbox, approval card,
and approval detail views.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When commenting on an issue, the reassign dropdown now defaults to the
last commenter who is not the current user, preventing accidental
self-reassignment. Falls back to the current issue assignee if no
other commenters exist.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add updateCompanyBrandingSchema restricting agent-updatable fields to name,
description, brandColor, and logoAssetId
- Update PATCH /api/companies/:companyId to allow CEO agents with branding-only
fields while keeping admin fields (status, budget, etc.) board-only
- Allow agents to GET /api/companies/:companyId for reading company info
- issuePrefix (company slug) remains protected — not in any update schema
- Document branding APIs in SKILL.md quick reference and api-reference.md
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Add a PR E2E workflow that runs the Playwright onboarding test on
every PR targeting master. Generates a minimal config file and lets
Playwright manage the server lifecycle. Runs in skip_llm mode so
no secrets are required.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The "Enable auto-merge" step runs unconditionally, even when the
lockfile didn't change and no PR exists. This causes the workflow
to fail with "lockfile PR was not found."
Use a step output to gate the auto-merge step so it only runs
when a PR was actually created or updated.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Blue dot indicator on Runs tab when there's an active run
- Run Now already switches to Runs tab (was done previously)
- LiveRunWidget shows streaming transcript in Runs tab for active runs
- Poll routine detail and runs list during active runs for real-time updates
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Larger cards (88px tall) with more breathing room (24px/56px gaps)
- Descriptive role labels (Chief Executive, Technology, etc.) instead of abbreviations
- SVG icon paths per role (star, terminal, globe, etc.) instead of text labels
- Keeps Pango-safe rendering (no emoji) while being visually closer to Warmth HTML
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Keep only Triggers, Runs, and Activity tabs per board request.
Cleaned up unused executionIssues query, IssueRow and ListTree imports.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The server rejected manual runs for any non-active routine. Now only
archived routines are blocked — paused routines can still be triggered
manually via "Run now".
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The DELETE /routine-triggers/:id endpoint returns 204 No Content, but
the API client unconditionally called res.json() on all responses,
causing a JSON parse error that surfaced as "API route not found".
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Change hour labels from "10:00 AM" to "10 AM" to avoid confusion with the separate minute selector
- Hide hour/minute selectors when "Every minute" preset is selected (no time config needed)
- Fix describeSchedule to work with updated hour label format
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The frontend generateReadmeFromSelection() was building an inline Mermaid
diagram for the org chart. The server already generates a PNG at
images/org-chart.png, so the preview should reference it the same way.
Removed dead mermaidId/mermaidEscape/generateOrgChartMermaid helpers.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add trash icon button to delete triggers (full stack: service, route, API client, UI)
- Fix pause/unpause bug where saving routine could revert status by excluding
status from the save payload (status is managed via dedicated pause/resume buttons)
- Add toast feedback for run, pause, and resume actions
- Auto-switch to Runs tab after triggering a manual run
- Add live update invalidation for routine/trigger/run activity events
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The MarkdownBody component now accepts an optional resolveImageSrc callback
that maps relative image paths (like images/org-chart.png) to base64 data URLs
from the portable file entries. This fixes the export README showing a broken
image instead of the org chart PNG.
Applied to both CompanyExport and CompanyImport preview panes.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add first-class --skills <list> option to `paperclipai company export`,
passing through to the existing service support for skill selection
- Remove broken `type: "url"` source branch from import command — the shared
schema and server only accept `inline | github`, so non-GitHub HTTP URLs
now error clearly instead of failing at validation
- Export isHttpUrl/isGithubUrl helpers for testability
- Add server tests for skills-filtered export (selected + fallback)
- Add CLI tests for URL detection helpers
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Remove Card wrappers and gray backgrounds from routine detail
- Use max-w-2xl container layout like issue detail page
- Add icons to tabs (Clock, Play, ListTree, Activity) matching issue tabs
- Make activity tab compact (single-line items with space-y-1.5)
- Create shared RunButton and PauseResumeButton components
- Build user-friendly ScheduleEditor with presets (hourly, daily, weekdays, weekly, monthly)
- Auto-detect timezone via Intl API instead of manual timezone input
- Use shared action buttons in both AgentDetail and RoutineDetail
- Replace bordered run history cards with compact divided list
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When skills are imported via skills.sh URLs or key-style imports
(org/repo/skill), the stored sourceType is now "skills_sh" with the
original skills.sh URL as sourceLocator, instead of "github" with the
resolved GitHub URL.
- Add "skills_sh" to CompanySkillSourceType and CompanySkillSourceBadge
- Track originalSkillsShUrl in parseSkillImportSourceInput
- Override sourceType/sourceLocator in importFromSource for skills.sh
- Handle skills_sh in key derivation, source info, update checks,
file reads, portability export, and UI badge rendering
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Clicking any routine row navigates to its detail/edit view
- Renamed "Open" to "Edit" in the three-dot context menu
- Added stopPropagation on toggle switch and dropdown cells
so interactive elements don't trigger row navigation
- Removed redundant Link wrapper on routine title
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove Card wrapper and header background (bg-muted/30)
- Remove uppercase tracking from header text
- Add hover state (hover:bg-accent/50) and border-b to rows
- Tighten cell padding to match issues list density
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add two new columns to the routines index table:
- Project column with a colored dot matching the project color
- Agent column with the agent icon and name
Moves the project info out of the Name cell subtitle into its own
dedicated column for better scannability.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add DELETE /api/companies/:companyId/skills/:skillId endpoint with same
permission model as other skill mutations. Deleting a skill removes it
from the DB, cleans up materialized runtime files, and automatically
strips it from any agent desiredSkills that reference it.
- Fix parseSkillImportSourceInput to detect skills.sh URLs
(e.g. https://skills.sh/org/repo/skill) and resolve them to the
underlying GitHub repo + skill slug, instead of fetching the HTML page.
- Add tests for skills.sh URL resolution with and without skill slug.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Align the routines list and detail editors with the issue-composer interaction model, and fix the scheduler's typed date comparison.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Lists all skills in a markdown table with name, description, and source.
GitHub and URL-sourced skills render as clickable links to their repository.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Inventories every file that touches the agentcompanies/v1-draft spec:
spec docs, shared types/validators, server services and routes, CLI
commands, UI pages/components/libraries, tests, and skills. Includes
a cross-reference table mapping spec concepts to implementation files.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Pi sometimes sends tool results as a direct array [{"type":"text","text":"..."}]
rather than wrapped in {"content": [...]}. Now handles both formats to
properly extract text content instead of showing raw JSON.
Pi returns tool results in format: {"content": [{"type": "text", "text": "..."}]}
Previously we were JSON.stringify-ing the whole object, showing as:
{"content":[{"type":"text","text":"..."}]}
Now we extract the actual text content for cleaner display.
The turn_end event includes toolResults array with toolName. Previously
the parsing only included toolCallId, now we also extract toolName so
the UI can display the correct tool name even when tool_result entries
arrive without a preceding tool_call.
When tool_result entries arrive without a matching tool_call, the transcript
was showing generic 'tool' as the name. Now pl-local parses toolName from
tool_execution_end events and passes it through, so the UI can display
the actual tool name (e.g., 'bash', 'Read', 'Ls') instead of 'tool'.
- Hoist PI_AGENT_SKILLS_DIR to module-level constant to avoid duplicate path computation
- Always create ~/.pi/agent/skills directory before early return in ensurePiSkillsInjected, so the path is valid when --skill flag is passed
- Add --skill ~/.pi/agent/skills to pi CLI invocation so it loads the paperclip skill
- Enable pi_local in the AgentConfigForm adapter type dropdown (was showing as coming soon)
Tasks are now loaded by default on the export page (unchecked and
folded as before). The "Load task files" / "Hide task files" button
is removed since it is no longer needed.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The tasks directory is now excluded from auto-expanded top-level
directories when the export page loads, keeping the tree cleaner.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Filter out projects with archivedAt set when building the export bundle,
so archived projects never appear in the exported package. Adds a warning
when archived projects are skipped.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Keep readable namespaced skill export folders while replacing managed company UUID segments with the company issue prefix for export-only paths.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The previous h-[calc(100dvh-6rem)] underestimated the vertical overhead
(breadcrumb, padding, worktree banner, button bar). Using h-full lets the
flex layout propagate the correct available height from <main>.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When a skill's source is "Paperclip workspace", clicking the label now
copies the absolute path to the managed skills workspace to the clipboard
and shows a toast confirmation.
- Add sourcePath field to CompanySkillDetail and CompanySkillListItem types
- Return managedRoot path as sourcePath from deriveSkillSourceInfo for
Paperclip workspace skills
- Make source label a clickable button in SkillPane detail view
Co-Authored-By: Paperclip <noreply@paperclip.ing>
New files created via the "+" button were not appearing in the file tree
because visibleFilePaths was derived solely from bundle API data. The
selection reset effect would also immediately undo the file selection.
Add pendingFiles state to track newly created files until they are saved
to disk. Include pending files in visibleFilePaths, guard the selection
reset effect, and clean up pending state after successful save.
Fixes PAP-563
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Reserve 8px scrollbar width at all times instead of expanding from 0 on
hover. The thumb stays transparent until hover so the scrollbar is
visually hidden but no longer causes a layout shift.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The useEffect that syncs selectedFile was resetting to an existing file
whenever the selected path wasn't in the bundle's on-disk file list.
This prevented selecting the entry file (e.g. AGENTS.md) when it didn't
yet exist on disk, even though it was visible in the file tree.
Allow selecting the currentEntryFile even when it has no on-disk file.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Expose adapter-discovered user-installed skills with provenance metadata, share persistent skill snapshot classification across local adapters, and render unmanaged skills as a read-only section in the agent skills UI.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add onClick handler to file row div in PackageFileTree so clicks
anywhere on the row select the file (not just the inner button)
- Replace "Add" button with compact "+" icon that reveals an inline
input with Create/Cancel actions
- Hide file name input until "+" is clicked to reduce visual clutter
- Validate new file paths: reject ".." path traversal segments
- Change placeholder from "docs/TOOLS.md" to "TOOLS.md"
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When the agent instructions tab is in "External" mode, the advanced
collapsible section now defaults to open so users don't have to manually
expand it to see the mode and path settings.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add upload, list, get content, and delete attachment endpoints
to the Key Endpoints table so agents know about the attachments API.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove hardcoded validTabs set and pass through whatever tab
segment is in the current URL. This makes tab preservation
work for all current and future agent tabs.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Set scrollbar width to 0 when not hovering so the track area
doesn't create a visible gutter. On hover, width expands to 8px
with the track/thumb colors.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Advanced collapsible now sits in its own row above the file browser/editor
- Increased spacing between form fields (gap-4 → gap-5, space-y-1 → space-y-1.5)
- Added more bottom padding (pb-6) to Advanced section for scroll room
- Increased inner spacing (space-y-4 → space-y-5) so mode/root path/entry file don't touch
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Comment out the Skills tab entry, view rendering, and breadcrumb
in AgentDetail so it's not visible. The code is preserved for
later re-enablement.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
No adapters currently support configuring compaction, so the info box
adds complexity without utility. Will revisit at the adapter level.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Only show the bootstrap prompt field on the agent configuration page
when editing an existing agent that already has a bootstrapPromptTemplate
value set. Label it as "(legacy)" with an amber notice recommending
migration to prompt template or instructions file. Hidden entirely
for new agent creation.
Closes PAP-536
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The native title attribute tooltip was not working reliably. Switched
to the project's Radix-based Tooltip component which provides instant,
styled tooltips on hover.
Fixes PAP-533
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The full version string was pushing the sidebar too wide. Now displays
just "v" with the full version (e.g. "v1.2.3") shown on hover via
title attribute, for both mobile and desktop sidebar layouts.
Fixes PAP-533
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The agent instructions tab was renamed from "prompts" to "instructions"
in 6b355e1, but the sidebar valid tabs set was not updated to include
the new name, causing tab preservation to fall back to dashboard.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Rename Prompts tab to Instructions (with backwards-compatible URL routing)
- Update header/subheader text to "Instructions Bundle" / "Configure your agent's behavior with instructions"
- Remove standalone legacy prompt warning banner; move warning to deprecated virtual file badge with tooltip
- Move mode, root path, and entry file controls into a collapsible "Advanced" section below the file browser
- Add help tooltips (?) for mode, root path, and entry file fields
- Show full root path in managed mode with copy-to-clipboard icon
- Reorder: root path now appears to the left of entry file
- Remove HEARTBEAT.md, SOUL.md, TOOLS.md shortcut buttons from file browser
- Add key prop to MarkdownEditor to ensure proper re-mount on file selection change
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When clicking a different agent in the sidebar, the current tab (prompts,
skills, configuration, budget, runs) is now preserved in the navigation
URL instead of always defaulting to dashboard. Falls back to default
agent URL for non-tab paths (e.g. specific run detail pages).
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Stabilize prompt file tree expansion state so the prompts editor no longer loops into maximum update depth when loading the bundle. Also replace bundle and file loading placeholders with skeleton UI.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Keep existing instructionsFilePath agents in external-bundle mode during edits, expose legacy promptTemplate as a deprecated virtual file, and reuse the shared PackageFileTree component in the Prompts view.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Address Greptile review feedback:
1. Wrap os.userInfo() in try/catch — it throws SystemError when the
current UID has no /etc/passwd entry (e.g. `docker run --user 1234`
with a minimal image). Falls back to process.env.HOME gracefully.
2. Add HOME to VOLATILE_ENV_KEY_EXACT so the discovery cache key is
not affected by the caller-supplied HOME vs the resolved HOME.
os.userInfo().homedir is constant for the process lifetime, so
HOME adds no useful cache differentiation.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When Paperclip's server is started via `runuser -u node` (common in
Docker/Fly.io deployments), the HOME environment variable retains the
parent process's value (e.g. /root) instead of the target user's home
directory (/home/node). This causes `opencode models` to miss provider
auth credentials stored under the actual user's home, resulting in
"Configured OpenCode model is unavailable" errors for providers that
require API keys (e.g. zai/zhipuai).
Fix: use `os.userInfo().homedir` (reads from /etc/passwd, not env) to
ensure the child process always sees the correct HOME, regardless of
how the server was launched.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Rearrange tabs to: Dashboard, Prompts, Skills, Configuration, Budget.
Move Prompt Template out of Configuration into a dedicated Prompts tab
with its own save/cancel flow and dirty tracking.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Decode HTML entities (e.g.  ) from title and description
before copying to clipboard, and trim trailing whitespace.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Adds a copy icon button to the left of the properties panel toggle
on the issue detail page. Clicking it copies a markdown representation
of the issue (identifier, title, description) to the clipboard and
shows a success toast. The icon briefly switches to a checkmark for
visual feedback.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Rearrange tabs to: Dashboard, Prompts, Skills, Configuration, Budget.
Move Prompt Template out of Configuration into a dedicated Prompts tab
with its own save/cancel flow and dirty tracking.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The execution workspace section wrapper div was rendered whenever a
project was selected, even when experimental workspaces were off.
The empty div's py-3 padding caused a visible layout bump. Now the
entire block only renders when currentProjectSupportsExecutionWorkspace
is true, and the redundant inner conditional is removed.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduce a singleton instance_settings store and experimental settings API, add the Experimental instance settings page, and gate execution workspace behavior behind the new enableIsolatedWorkspaces flag.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Same styling fix as NewIssueDialog — removes the rounded-md border
from the workspace toggle in the issue detail view.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Removes the rounded border around the execution workspace toggle section
and increases top/bottom padding for better visual spacing.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The zod schema validates archivedAt as a datetime string, but Drizzle's
timestamp column expects a Date object. The string was passed directly to
db.update(), causing a 500 error. Now we convert the string to a Date
in the route handler before calling the service.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When dragging or pasting an image into a markdown editor field, the cursor
would end up right next to the image making it hard to continue typing.
Now inserts two newlines after the image so a new paragraph is ready.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Headers were butting up against previous paragraphs too closely. Changed
rendered markdown header selectors from :where() to direct element selectors
to increase CSS specificity and beat Tailwind prose defaults. Bumped
margin-top from 1.15rem to 1.75rem. Also added top margins to MDXEditor
headers (h1: 1.4em, h2: 1.3em, h3: 1.2em) which previously had none.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Swap the browser alert dialog for an in-page confirm/cancel button pair.
Shows a loading spinner while the archive request is in flight, then the
redirect and toast (from prior commit) handle the rest.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The sticky float-right save/cancel bar was rendering (invisible via
opacity-0) on all tabs including runs, causing it to push the runs
layout content. Now only rendered when showConfigActionBar is true.
Also reverts the negative margin workaround from the previous attempt.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When no goals are linked, hide the "None" label and just show the
"+ Goal" button. When goals exist, add left margin to the button so
it doesn't crowd the goal badges.
Closes PAP-522
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously archiving a project silently navigated to /projects with no
feedback. Now it navigates to /dashboard and shows a success toast for
both archive and unarchive actions.
Closes PAP-521
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use negative right margin to counteract the Layout container padding,
giving the runs detail panel more horizontal space especially on
smaller screens.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Rename "List" tab to "Issues" and reorder: Issues, Overview, Configuration
- Cache the last active tab per project in localStorage
- On revisit, restore the cached tab instead of always defaulting to Issues
Closes PAP-520
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Org chart now uses a Mermaid flowchart (graph TD) instead of a
standalone SVG file — GitHub and the preview both render it natively
- Removed SVG generation code, layout algorithm, and image resolution
- Removed images/org-chart.svg from export output
- Simplified ExportPreviewPane (no more SVG/data-URI handling)
- Both server and client README generators produce Mermaid diagrams
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The export page now syncs the selected file with the URL path, e.g.
/PAP/company/export/files/agents/cmo/AGENTS.md. Navigating to such a
URL directly selects and reveals the file in the tree. Browser
back/forward navigation is supported without page refreshes.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Replace break-all with <wbr> hints after / and - characters so paths
break at directory and word boundaries instead of mid-word
- Use overflow-wrap: anywhere as fallback for very long segments
- Apply natural breaking to the workspace name link as well
- Rename CopyablePath to CopyableValue with optional mono prop for
better semantic clarity
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- README.md now regenerates in real-time when files are checked/unchecked
in the export file tree, so counts and tables reflect the actual selection
- SVG image references in markdown (e.g. images/org-chart.svg) resolve to
inline data URIs so the org chart renders in the README preview
- Fixes issue where unchecked tasks/projects were still counted in README
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Replace truncated paths with wrapping text (break-all) so full paths
are visible
- Add CopyablePath component with a copy icon that appears on hover and
shows a green checkmark after copying
- Apply to all workspace paths: cwd, branch name, repo URL, and the
project primary workspace fallback
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Display the absolute cwd path, branch name, and repo URL for the current
execution workspace in the issue properties panel. When no execution
workspace is active yet, show the project's primary workspace path as
a fallback.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Remove "Operator branch" and "Agent default" options from the workspace
mode dropdown in both NewIssueDialog and IssueProperties, keeping only
"Project default", "New isolated workspace", and "Reuse existing workspace"
- Deduplicate reusable workspaces by space identity (cwd path) so the
"choose existing workspace" dropdown shows unique worktrees instead of
duplicate entries from multiple issues sharing the same local folder
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add inline SVG rendering for .svg files in ExportPreviewPane
- Update Getting Started to use simpler `company import` syntax
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Adds auto-generated README.md with company summary, agent table, project
list, and getting-started instructions. Includes an SVG org chart image
in images/org-chart.svg using the same layout algorithm as the UI.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Use flex layout so the canvas fills remaining space after the button bar,
instead of a fixed viewport calc that didn't account for button height.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
clearRepoWorkspace was calling updateWorkspace to null out the repo
even when there was no local folder, leaving an empty workspace.
Now falls through to persistCodebase which correctly removes the
entire workspace when both cwd and repoUrl would be null.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The invalidateProject() in ProjectProperties only invalidated
queryKeys.projects.detail(project.id) (UUID), but the parent
ProjectDetail query uses routeProjectRef which is typically the
URL key (e.g. "openclaw-testing"). This meant mutations succeeded
but the parent query was never refetched — the UI only updated on
page refresh.
Now also invalidates via project.urlKey so both UUID and URL-key
based query caches are cleared.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Allow saving empty values to clear repo URL or local folder from an existing workspace
- submitLocalWorkspace/submitRepoWorkspace now handle empty input as a "clear" operation
- Save button is only disabled for empty input when there's no existing workspace to clear
- removeWorkspace.onSuccess now resets form state (matching create/update handlers)
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove the org/repo namespace line from both the company skills sidebar
and agent skills tab — cleaner to show just the skill name with source
icon.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Per feedback, the detail page looks cleaner without the org/repo
namespace line above the skill name. The icon remains on the name row.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The namespace was appearing side-by-side with the skill name because
they were in the same flex row. Restructured the layout so the
namespace appears on its own line above, with the icon and skill name
on the row below it.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
On the skill detail page, the namespace (e.g. org/repo) now appears
above the skill name in small monospace text, and the source icon is
placed on the same row as the skill name rather than in the metadata.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
In both the company skills list and agent skills tab, the skill key
(e.g. org/repo/skill) is now split so only the namespace portion
(org/repo) appears above the skill name, rather than the full key
below it. Non-namespaced skills show no namespace line.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- When no goals are linked, just show the "+ Goal" button without
displaying "None" text
- Add left margin to the "+ Goal" button when goals exist above it
for better visual spacing
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The archive mutation lacked user feedback:
- No toast notification on success/failure
- Navigated to /projects instead of /dashboard after archiving
- No error handling if the API call failed
Added success/error toasts using the existing ToastContext, navigate
to /dashboard after archiving (matching user expectations), and error
toast on failure.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The updateWorkspace mutation's onSuccess handler only invalidated the
project query but didn't reset the form state (close the edit mode,
clear inputs). This made it look like Save did nothing when editing an
existing workspace. Now matches createWorkspace's onSuccess behavior.
Also added updateWorkspace.isPending to the Save button disabled state
for both local folder and repo inputs.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove unnecessary `as any` casts on capability strings (now valid
PluginCapability members) and add company-membership guards to match
production behavior in plugin-host-services.ts.
Wire issue document list/get/upsert/delete operations through the
JSON-RPC protocol so plugins can manage issue documents with the same
capabilities available via the REST API.
Fixes#940
* public-gh/master: (51 commits)
Use attachment-size limit for company logos
Address Greptile company logo feedback
Drop lockfile from PR branch
Use asset-backed company logos
fix: use appType "custom" for Vite dev server so worktree branding is applied
docs: fix documentation drift — adapters, plugins, tech stack
docs: update documentation for accuracy after plugin system launch
chore: ignore superset artifacts
Dark theme for CodeMirror code blocks in MDXEditor
Remove duplicate @paperclipai/adapter-openclaw-gateway in server/package.json
Fix code block styles with robust prose overrides
Add Docker setup for untrusted PR review in isolated containers
Fix org chart canvas height to fit viewport without scrolling
Add doc-maintenance skill for periodic documentation accuracy audits
Fix sidebar scrollbar: hide track background when not hovering
Restyle markdown code blocks: dark background, smaller font, compact padding
Add archive project button and filter archived projects from selectors
fix: address review feedback — subscription cleanup, filter nullability, stale diagram
fix: wire plugin event subscriptions from worker to host
fix(ui): hide scrollbar track background when sidebar is not hovered
...
# Conflicts:
# packages/db/src/migrations/meta/0030_snapshot.json
# packages/db/src/migrations/meta/_journal.json
Required/built-in Paperclip skills are now shown in a dedicated
"Required by Paperclip" section at the bottom of the agent skills tab,
with checkboxes that are checked and disabled. Optional skills remain
in the main section above.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Extract all Anthropic credential/API logic into claude-local/src/server/quota.ts
- Extract all OpenAI/WHAM credential/API logic into codex-local/src/server/quota.ts
- Add optional getQuotaWindows() to ServerAdapterModule in adapter-utils
- Rewrite quota-windows.ts as a 29-line thin aggregator with zero provider knowledge
- Wire getQuotaWindows into adapter registry for claude-local and codex-local
- Add 47 unit tests covering toPercent, secondsToWindowLabel, WHAM normalization,
readClaudeToken, readCodexToken, fetchClaudeQuota, fetchCodexQuota, fetchWithTimeout
- Add 8 unit tests covering parseDateRange validation and byProvider pro-rata math
Adding a third provider now requires only touching that provider's adapter.
- add byAgentModel endpoint and expandable per-agent model sub-rows in the spend tab
- validate date range inputs with isNaN + badRequest to return HTTP 400 on bad input
- move CostByProject from a local api/costs.ts definition into packages/shared types
- gate providerData query on mainTab === providers, consistent with weekData/windowData/quotaData
- fix byProject range filter from finishedAt to startedAt, consistent with byProvider runs query
- fix WHAM used_percent threshold from <= 1 to < 1 to avoid misclassifying 1% usage as 100%
- replace inline opacity style with tailwind bg-primary/85 class in ProviderQuotaCard
- reset expandedAgents set when company or date range changes
- sort agent model sub-rows by cost descending in ui memo
- add company existence check on quota-windows route to guard against
sentinel and forged company IDs (was a no-op assertCompanyAccess)
- fix useDateRange minuteTick memo frozen at mount; realign interval to
next calendar minute boundary via setTimeout + intervalRef pattern
- fix midnight timer in Costs.tsx to use stable [] dep and
self-scheduling todayTimerRef to avoid StrictMode double-invoke
- return null for rolling window rows with no DB data instead of
rendering $0.00 / 0 tok false zeros
- fix secondsToWindowLabel to handle windows >168h with actual day count
instead of silently falling back to 7d
- fix byProvider.get(p) non-null assertion to use ?? [] fallback
merge Usage page into Costs as two tabs ('Spend' and 'Providers'),
extract shared date-range logic to useDateRange() hook, delete /usage
route and sidebar entry, fix quota-windows bugs from prior review
reads local claude and codex auth files server-side, calls provider
quota apis (anthropic oauth usage, chatgpt wham/usage), and surfaces
live usedPercent per window in ProviderQuotaCard with threshold fill colors
adds a new /usage page that lets board operators see how much each ai
provider is consuming across any date window, with per-model breakdowns,
rolling 5h/24h/7d burn windows, weekly budget bars, and a deficit notch
when projected spend is on track to exceed the monthly budget.
- new GET /companies/:id/costs/by-provider endpoint aggregates cost events
by provider + model with pro-rated billing type splits from heartbeat runs
- new GET /companies/:id/costs/window-spend endpoint returns rolling window
spend (5h, 24h, 7d) per provider with no schema changes
- QuotaBar: reusable boxed-border progress bar with green/yellow/red
threshold fill colors and optional deficit notch
- ProviderQuotaCard: per-provider card showing budget allocation bars,
rolling windows, subscription usage, and model breakdown with token/cost
share overlays
- Usage page: date preset toggles (mtd, 7d, 30d, ytd, all, custom),
provider tabs, 30s polling plus ws invalidation on cost_event
- custom date range blocks queries until both dates are selected and
treats boundaries as local-time (not utc midnight) so full days are
included regardless of timezone
- query key to timestamp is floored to the nearest minute to prevent
cache churn on every 30s refetch tick
- Added hideInstructionsFile prop to AdapterConfigFieldsProps
- All adapter config-fields now conditionally hide the instructions file
field when hideInstructionsFile is set (used during import since the
AGENTS.md is automatically set as promptTemplate)
- Import adapter config panel now renders ClaudeLocalAdvancedFields
(Chrome, skip permissions, max turns) when claude_local is selected
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When importing a company, users can now choose the adapter type for each
imported agent. Defaults to the current company CEO's adapter type (or
claude_local if none). Includes an expandable "configure adapter" section
per agent that renders the adapter-specific config fields.
- Added adapterOverrides to import request schema and types
- Built AdapterPickerList UI component in CompanyImport.tsx
- Backend applies adapter overrides when creating/updating agents
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The /company/import page now defaults the target dropdown to "Create new
company" instead of the current company. The existing company option is
still available in the dropdown.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Move the rename indicator (→ newName) in the file tree to only
display on the parent directory node, not on the individual file.
The preview header still shows the rename when viewing the file.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Move "Import n files" button to right side of its container
- Show "→ newName" rename indicator next to files/directories in the
file tree when an agent or project is being renamed on import
- Show "→ newName" rename indicator in the file preview header when
viewing a file that will be renamed
- Uses cyan color to distinguish rename info from action badges
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Move the "Import n files" button from the sticky header bar to below
the renames confirmation panel, so the user reviews renames first
before importing.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Remove COMPANY.md from renames panel; just uncheck it silently in
the file tree when importing to existing company
- Rename panel from "Conflicts to resolve" to "Renames"
- Add "skip" button on the left and "confirm rename" button on the
right of each rename row
- Confirmed renames show a green checkmark and green-tinted row
- Skipped items gray out and uncheck the file in the tree
- Un-confirmed renames still proceed with the rename by default
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Remove the collision strategy dropdown; always default to "rename"
- Add a "Conflicts to resolve" chores list above the package file tree
showing each collision with editable rename fields (oldname → newname)
- Default rename uses source folder prefix (e.g. gstack-CEO)
- Per-item "skip" button that syncs with file tree checkboxes
- COMPANY.md defaults to skip when importing to an existing company
- Add nameOverrides support to API types and server so user-edited
renames are passed through to the import
Co-Authored-By: Paperclip <noreply@paperclip.ing>
We don't support regular updates to agents from GitHub sources yet,
so the "not pinned to a commit SHA" warning is misleading and unnecessary.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extract duplicated file tree types, helpers (buildFileTree, countFiles,
collectAllPaths, parseFrontmatter), and visual tree component into a
shared PackageFileTree component. Both import and export pages now use
the same underlying tree with consistent alignment and styling.
Import-specific behavior (action badges, unchecked opacity) is handled
via renderFileExtra and fileRowClassName props. Also removes the file
count subtitle from the import sidebar to match the export page.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When importing into an existing company, files with "update" action (conflicts)
now have their parent directories auto-expanded so users immediately see what
will be overwritten. Additionally, server-side warnings are generated for any
agent or project that will be overwritten by the import.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- update: blue → yellow (amber)
- overwrite/replace: added as red
- create (green) and skip (gray) unchanged
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
No collisions are possible when the target is a new company, so the
dropdown is unnecessary. The grid layout also adjusts to single-column
when only the target field is shown.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Change file row outer gap from gap-2 (8px) to gap-1 (4px) to match
the directory row grid gap-x-1, so file and folder icons line up
vertically.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When users check/uncheck files in the export preview, the .paperclip.yaml
now dynamically filters its agents/projects/tasks sections to only include
entries whose corresponding files are checked. This applies to both the
preview pane and the downloaded tar archive.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Clicking a skill pill in the frontmatter card now navigates to the
corresponding skills/<slug>/SKILL.md file in the export tree, expanding
parent directories as needed. No page reload required.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Move paddingLeft from inner label to outer grid div on directory rows
so folders align with files and the search field
- Remove "N files in rootPath" subtitle under Package files header
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Show only 10 task entries at a time with a "Show more issues" button.
Checked/selected tasks are always pinned visible regardless of the page
limit. Search still works across all issues — matched results are pinned
and the load-more button is hidden during search so all matches show.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove "N notes" indicator from the top bar
- Hide terminated agent messages entirely instead of showing as notes
- Style warnings as a rounded box with side borders and more margin
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Vite's "spa" appType adds its own SPA fallback middleware that serves
index.html directly, bypassing the custom catch-all route that calls
applyUiBranding(). Changing to "custom" lets our route handle HTML
serving, which injects the worktree-colored favicon and banner meta tags.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address Greptile review feedback:
- Show "Please fill in all required fields." instead of silently
returning when form is submitted with missing fields
- Remove pointer-events-none so keyboard users can reach the
button and receive the same validation feedback
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- README: mark plugin system as shipped in roadmap
- SPEC: update adapter table with openclaw_gateway, gemini-local, hermes_local
- SPEC: update plugin architecture section to reflect shipped status
- Add .doc-review-cursor for future maintenance runs
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Move all tasks to top-level tasks/ folder (no longer nested under
projects/slug/tasks/). The project slug is still in the frontmatter
for association.
- Search auto-expands parent dirs of matched files so matches are
always visible in the tree
- Restores previous expansion state when search is cleared
- All files already loaded in memory — search works across everything
with no pagination limit
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add `files` and `manifest` to CompanyPortabilityPreviewResult so the
import UI can show actual file contents and metadata
- Rewrite import preview as a file/folder tree (matching export page
design language) with per-file checkboxes to include/exclude items
- Show action badges (create/update/skip) on each file based on the
import plan, with unchecked files dimmed and badged as "skip"
- Add rich frontmatter preview: clicking a file shows parsed frontmatter
as structured data (name, title, reportsTo, skills) plus markdown body
- Include skills count in the sidebar summary
- Update import button to show dynamic file count that updates on
check/uncheck
- Both /tree/ and /blob/ GitHub URLs already supported by backend
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Separate terminated agent messages from warnings into info notes
(shown with subtle styling instead of amber warning banners)
- Clean up warning banner styles for dark mode compatibility
(use amber-500/20 borders and amber-500/5 backgrounds)
- Parse YAML frontmatter in markdown files and render as structured
data cards showing name, title, reportsTo, skills etc.
- Apply same warning style cleanup to import page
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add /:company/company/export page with file tree, checkboxes for
per-file selection, and read-only preview pane (skills-style layout)
- Add /:company/company/import page with source form (GitHub/URL/local),
target/collision settings, preview tree with action badges, and detail pane
- Add Import/Export buttons to the Org Chart page header
- Replace import/export sections in CompanySettings with redirect links
- Clean up ~800 lines of dead code from CompanySettings
- Register new routes in App.tsx
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The code blocks users see in issue documents are rendered by CodeMirror
(via MDXEditor's codeMirrorPlugin), not by MarkdownBody. MDXEditor
bundles cm6-theme-basic-light which gives them a white background.
Added dark overrides for all CodeMirror elements:
- .cm-editor: dark background (#1e1e2e), light text (#cdd6f4)
- .cm-gutters: darker gutter with muted line numbers
- .cm-activeLine, .cm-selectionBackground: subtle dark highlights
- .cm-cursor: light cursor for visibility
- Language selector dropdown: dark-themed to match
- Reduced pre padding to 0 since CodeMirror handles its own spacing
Uses \!important to beat CodeMirror's programmatically-injected theme
styles (EditorView.theme generates high-specificity scoped selectors).
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Previous attempt was being overridden by Tailwind prose/prose-invert
CSS variables. This fix:
- Overrides --tw-prose-pre-bg and --tw-prose-invert-pre-bg CSS variables
on .paperclip-markdown to force dark background in both modes
- Uses .paperclip-markdown pre with \!important for bulletproof overrides
- Removes conflicting prose-pre: utility classes from MarkdownBody
- Adds explicit pre code reset (inherit color/size, no background)
- Verified visually with Playwright at desktop and mobile viewports
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Adds a dedicated Docker environment for reviewing untrusted pull requests
with codex/claude, keeping CLI auth state in volumes and using a separate
scratch workspace for PR checkouts.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The height calc subtracted only 4rem but the actual overhead is ~6rem
(3rem breadcrumb bar + 3rem main padding). Also use dvh for better
mobile support.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Skill detects documentation drift by scanning git history since last review,
cross-referencing shipped features against README, SPEC, and PRODUCT docs,
and opening PRs with minimal fixes. Includes audit checklist and section map
references.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The scrollbar track background was still visible as a colored "well" even
when the thumb was hidden. Now both track and thumb are fully transparent
by default, only appearing on container hover.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Switch code block background from transparent accent to dark (#1e1e2e) with
light text (#cdd6f4) for better readability in both light and dark modes
- Reduce code font size from 0.84em to 0.78em
- Compact padding and margins on pre blocks
- Hide MDXEditor code block toolbar by default, show on hover/focus to prevent
overlap with code content on mobile
- Use horizontal scroll instead of word-wrap for code blocks to preserve formatting
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Add scopedBus.clear() in dispose() to prevent subscription accumulation
on worker crash/restart cycles
- Use two-arg subscribe() overload when filter is null instead of passing
empty object; fix filter type to include null
- Update ASCII flow diagram: onEvent is a notification, not request/response
Move inline test cases from promptfooconfig.yaml into separate files
organized by category (core.yaml, governance.yaml). Main config now
uses file://tests/*.yaml glob pattern per promptfoo best practices.
This makes it easier to add new test categories without bloating the
main config, and lets contributors add cases by dropping new YAML
files into tests/.
Plugin workers register event handlers via `ctx.events.on()` in the SDK,
but these subscriptions were never forwarded to the host process. The host
sends events via `notifyWorker("onEvent", ...)` which produces a JSON-RPC
notification (no `id`), but the worker only dispatched `onEvent` as a
request handler — notifications were silently dropped.
Changes:
- Add `events.subscribe` RPC method so workers can register subscriptions
on the host-side event bus during setup
- Handle `onEvent` notifications in the worker notification dispatcher
(previously only `agents.sessions.event` was handled)
- Add `events.subscribe` to HostServices interface, capability map, and
host client handler
- Add `subscribe` handler in host services that registers on the scoped
plugin event bus and forwards matched events to the worker
Skill guides users through creating Agent Companies spec-conformant
packages, either from scratch (with interview-driven hiring plan) or
by analyzing an existing git repo and wrapping its skills/agents.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The scrollbar-auto-hide utility was only hiding the thumb but left the
track background visible, creating a visible "well" even when idle.
Now both track and thumb are transparent by default, appearing only on
container hover.
Fixes PAP-374
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Added packagePath to plugin loader for improved manifest handling.
- Refactored GlobalToolbarPlugins for better slot and launcher management in BreadcrumbBar.
- Updated launcher trigger styles for globalToolbarButton.
Address Greptile review feedback:
- Log plugin event handler errors via logger.warn instead of
silently discarding the emit() result
- Warn if setPluginEventBus is called more than once
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The plugin event bus accepts subscriptions for core events like
issue.created but nothing emits them. This adds a bridge in
logActivity() so every domain action that's already logged also
fires a PluginEvent to subscribing plugins.
Uses a module-level setter (same pattern as publishLiveEvent)
to avoid threading the bus through all route handlers. Only
actions matching PLUGIN_EVENT_TYPES are forwarded.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* public-gh/master:
Drop lockfile from watcher change
Tighten plugin dev file watching
Fix plugin smoke example typecheck
Fix plugin dev watcher and migration snapshot
Clarify plugin authoring and external dev workflow
Expand kitchen sink plugin demos
fix: set AGENT_HOME env var for agent processes
Add kitchen sink plugin example
Simplify plugin runtime and cleanup lifecycle
Add plugin framework and settings UI
# Conflicts:
# packages/db/src/migrations/meta/0029_snapshot.json
# packages/db/src/migrations/meta/_journal.json
Matches the pattern in codex-local and cursor-local adapters,
giving operators consistent feedback about whether instructions
were actually loaded.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
realpathSync resolves symlinks and normalizes case, preventing
double-loading the same .env file when paths differ only by
symlink indirection or filesystem case.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Escape `[` and `]` in filenames to prevent malformed markdown when
attaching images. Use `\n\n` instead of `\n` so the image renders
as its own paragraph instead of inline with preceding text.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The $AGENT_HOME environment variable was referenced by skills (e.g.
para-memory-files) but never actually set, causing runtime errors like
"/HEARTBEAT.md: No such file or directory" when agents tried to resolve
paths relative to their home directory.
Add agentHome to the paperclipWorkspace context in the heartbeat service
and propagate it as the AGENT_HOME env var in all local adapters.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The submit button's `disabled` attribute prevented browsers from firing
implicit form submission (Enter key) per HTML spec. Move the canSubmit
guard into the onSubmit handler and use aria-disabled + visual styles
instead, so Enter works when fields are filled.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Pin hermes-paperclip-adapter to exact version 0.1.1 (was ^0.1.0).
Avoids auto-pulling potentially breaking patches from a 0.x package.
- Enable supportsLocalAgentJwt (was false). The adapter uses
buildPaperclipEnv which passes the JWT to the child process,
matching the pattern of all other local adapters.
Resolve conflicts by keeping the issue-documents work alongside upstream heartbeat-context, worktree branding, and adapter runtime updates.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The server only loads environment variables from .paperclip/.env, which is
not the standard location users expect. When DATABASE_URL is set in a .env
file in the project root (cwd), it is silently ignored, requiring users to
manually export the variable.
Add a fallback that loads cwd/.env after .paperclip/.env with override:false,
so the Paperclip-specific env file always takes precedence but standard .env
files in the project root are also picked up.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- Make company_boundary test adversarial with cross-company stimulus
- Replace fragile not-contains:retry with targeted JS assertion
- Replace not-contains:create with not-contains:POST /api/companies
- Pin promptfoo to 0.103.3 for reproducible eval runs
- Fix npm -> pnpm in README prerequisites
- Add trailing newline to system prompt
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Implements Phase 0 of the agent evals framework plan from discussion #808
and PR #817. Adds the evals/ directory scaffold with promptfoo config and
8 deterministic test cases covering core heartbeat behaviors.
Test cases:
- core.assignment_pickup: picks in_progress before todo
- core.progress_update: posts status comment before exiting
- core.blocked_reporting: sets blocked status with explanation
- governance.approval_required: reviews approval before acting
- governance.company_boundary: refuses cross-company actions
- core.no_work_exit: exits cleanly with no assignments
- core.checkout_before_work: always checks out before modifying
- core.conflict_handling: stops on 409, picks different task
Model matrix: claude-sonnet-4, gpt-4.1, codex-5.4, gemini-2.5-pro via
OpenRouter. Run with `pnpm evals:smoke`.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Paperclip <noreply@paperclip.ing>
* public-gh/master: (33 commits)
fix: align embedded postgres ctor types with initdbFlags usage
docs: add dated plan naming rule and align workspace plan
Expand workspace plan for migration and cloud execution
Add workspace product model plan
docs: add token optimization plan
docs: organize plans into doc/plans with date prefixes
fix: keep runtime skills scoped to ./skills
fix: prefer .agents skills and repair codex symlink targets\n\nCo-Authored-By: Paperclip <noreply@paperclip.ing>
Change sidebar Documentation link to external docs.paperclip.ing
Fix local-cli skill install for moved .agents skills
docs: update PRODUCT.md and add 2026-03-13 features plan
feat(worktree): add worktree:cleanup command, env var defaults, and auto-prefix
fix: resolve type errors in process-lost-reaper PR
fix(heartbeat): prevent false process_lost failures on queued and non-child-process runs
Revert "Merge pull request #707 from paperclipai/nm/premerge-lockfile-refresh"
ci: refresh pnpm lockfile before merge
fix(docker): include gemini adapter manifest in deps stage
chore(lockfile): refresh pnpm-lock.yaml
Raise default max turns to 300
Drop pnpm lockfile from PR
...
When creating a new issue with a pre-filled assignee or project (e.g. from
a project page), Tab from the title field now skips over fields that already
have values, going directly to the next empty field or description.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When creating a new issue with a pre-filled assignee or project (e.g. from
a project page), Tab from the title field now skips over fields that already
have values, going directly to the next empty field or description.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Move plans from doc/plan/ into doc/plans/ and add YYYY-MM-DD date
prefixes to all undated plan files based on document headers or
earliest git commit dates.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The sidebar Documentation links were pointing to an internal /docs route.
Updated both mobile and desktop sidebar instances to link to
https://docs.paperclip.ing/ in a new tab instead.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* 'master' of github.com-dotta:paperclipai/paperclip:
docs: update PRODUCT.md and add 2026-03-13 features plan
feat(worktree): add worktree:cleanup command, env var defaults, and auto-prefix
- Add `worktree:cleanup <name>` command that safely removes a worktree,
its branch, and instance data. Idempotent and safe by default — refuses
to delete branches with unique unmerged commits or worktrees with
uncommitted changes unless --force is passed.
- Support PAPERCLIP_WORKTREES_DIR env var as default for --home across
all worktree commands.
- Support PAPERCLIP_WORKTREE_START_POINT env var as default for
--start-point on worktree:make.
- Auto-prefix worktree names with "paperclip-" if not already present,
so `worktree:make subissues` creates ~/paperclip-subissues.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix malformed try/catch/finally blocks in heartbeat executeRun
- Declare activeRunExecutions Set to track in-flight runs
- Add resumeQueuedRuns function and export from heartbeat service
- Add initdbFlags to EmbeddedPostgresCtor type
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Published hermes-paperclip-adapter@0.1.0 to npm registry
- Replaced github:NousResearch/hermes-paperclip-adapter with
hermes-paperclip-adapter ^0.1.0 (proper semver, reproducible builds)
- Updated imports from @nousresearch/paperclip-adapter-hermes to
hermes-paperclip-adapter
- Wired in hermesSessionCodec for structured session validation
Addresses both review items from greptile-apps:
1. Unpinned GitHub dependency → now a proper npm package with semver
2. Missing sessionCodec → now imported and registered
- Pin @nousresearch/paperclip-adapter-hermes to v0.1.0 tag for
reproducible builds and supply-chain safety
- Import and wire hermesSessionCodec into the adapter registration
for structured session parameter validation (matching claude_local,
codex_local, and other adapters that support session persistence)
- reapOrphanedRuns() now only scans running runs; queued runs are
legitimately absent from runningProcesses (waiting on concurrency
limits or issue locks) so including them caused false process_lost
failures (closes#90)
- Add module-level activeRunExecutions set so non-child-process adapters
(http, openclaw) are protected from the reaper during execution
- Add resumeQueuedRuns() to restart persisted queued runs after a server
restart, called at startup and each periodic tick
- Add outer catch in executeRun() so setup failures (ensureRuntimeState,
resolveWorkspaceForRun, etc.) are recorded as failed runs instead of
leaving them stuck in running state
- Guard resumeQueuedRuns() against paused/terminated/pending_approval agents
- Increase opencode models discovery timeout from 20s to 45s
When creating a new issue with a pre-filled assignee or project (e.g. from
a project page), Tab from the title field now skips over fields that already
have values, going directly to the next empty field or description.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
The sidebar Documentation links were pointing to an internal /docs route.
Updated both mobile and desktop sidebar instances to link to
https://docs.paperclip.ing/ in a new tab instead.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* public-gh/master:
Drop pnpm lockfile from PR
Update ui/src/components/OnboardingWizard.tsx
Update ui/src/components/OnboardingWizard.tsx
Fix onboarding manual debug JSX
Improve onboarding defaults and issue goal fallback
Simplify adapter environment check: animate pass, show debug only on fail
Show Claude Code and Codex as recommended, collapse other adapter types
Animate onboarding layout when switching between Company and Agent steps
Make onboarding wizard steps clickable tabs for easier dev navigation
Add agent chat architecture plan
Style tweaks for onboarding wizard step 1
Add direct onboarding routes
- When the adapter probe passes, show a clean animated "Passed" badge with
a checkmark icon instead of verbose check details
- Only show the "Manual debug" section when the test fails, reducing noise
for successful flows
Co-Authored-By: Paperclip <noreply@paperclip.ing>
In the onboarding wizard step 2 ("Create your first agent"), Claude Code and
Codex are now shown prominently as recommended options. Other adapter types
(OpenCode, Pi, Cursor, OpenClaw Gateway) are hidden behind a collapsible
"More Agent Adapter Types" toggle to reduce visual noise for new users.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
When on the Company step (step 1), the form takes up the left half with
ASCII art on the right. Switching to Agent or any other step smoothly
animates the form to full width while fading out the ASCII art panel.
Switching back reverses the animation.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace the progress bar dots with labeled tab buttons (Company, Agent,
Task, Launch) that allow clicking directly to any step. This makes it
easy to debug/preview individual onboarding screens without stepping
through the full flow.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Set animation panel (right half) background to #1d1d1d
- Add more margin above Company name form label
- Make form labels white when input is focused or has value
using Tailwind group/focus-within pattern
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Move useMemo and derived state above early returns so hooks are always
called in the same order. Simplify the description to plain English
and change toggle button labels to "Enable Timer Heartbeat" /
"Disable Timer Heartbeat" for clarity.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Group agents by company with a single card per company and dense
inline rows instead of one card per agent. Replaces the three stat
cards with a slim inline summary. Each row shows status badge, linked
agent name, role, interval, last heartbeat time, a config link icon,
and an enable/disable button.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
* public-gh/master:
Default Gemini adapter to yolo mode and add API access prompt note
fix: remove Cmd+1..9 company-switch shortcut
fix(ui): prevent IME composition Enter from moving focus in new issue title
fix(cli): add restart hint after allowed-hostname change
docs: remove obsolete TODO for CONTRIBUTING.md
fix: default dangerouslySkipPermissions to true for unattended agents
fix: route heartbeat cost recording through costService
Show issue creator in properties sidebar
Add a "Worktree CLI Reference" subsection to doc/DEVELOPING.md with
complete option tables and examples for worktree init, worktree:make,
and worktree env subcommands.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Gemini CLI only registers run_shell_command in --approval-mode yolo.
Non-yolo modes don't expose it at all, making Paperclip API calls
impossible. Always pass --approval-mode yolo and remove the now-unused
policy engine code, approval mode config, and UI toggles.
Add a "Paperclip API access note" to the prompt with curl examples
via run_shell_command, since the universal SKILL.md is tool-agnostic.
Also extract structured question events from Gemini assistant messages
to support interactive approval flows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* 'master' of github.com-dotta:paperclipai/paperclip:
Tighten transcript label styling
Fix env-sensitive worktree and runtime config tests
Refine executed command row centering
Tighten live run transcript streaming and stdout
Center collapsed command group rows
Refine collapsed command failure styling
Tighten command transcript rows and dashboard card
Polish transcript event widgets
Refine transcript chrome and labels
fix: remove paperclip property from OpenClaw Gateway agent params
Add a run transcript UX fixture lab
Humanize run transcripts across run detail and live surfaces
fix(adapters/gemini-local): address PR review feedback
fix(adapters/gemini-local): inject skills into ~/.gemini/ instead of tmpdir
fix(adapters/gemini-local): downgrade missing API key to info level
feat(adapters/gemini-local): add auth detection, turn-limit handling, sandbox, and approval modes
fix(adapters/gemini-local): address PR review feedback for skills and formatting
feat(adapters): add Gemini CLI local adapter support
# Conflicts:
# cli/src/__tests__/worktree.test.ts
Long unbroken text (stack traces, file paths, URLs) in issue descriptions
could overflow the container. Add break-words to MarkdownBody and
overflow-hidden to InlineEditor display wrapper.
Fixes PAP-476
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Ensure worktree init writes PAPERCLIP_AGENT_JWT_SECRET into the new .paperclip/.env when the source instance already has a usable secret loaded or configured. Also harden the affected integration tests against shell env leakage and full-suite timeout pressure.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
On macOS, `initdb` defaults to SQL_ASCII encoding because it infers
locale from the system environment. When `ensurePostgresDatabase()`
creates a database without specifying encoding, the new database
inherits SQL_ASCII from the cluster. This causes string functions like
`left()` to operate on bytes instead of characters, producing invalid
UTF-8 when multi-byte characters are truncated.
Two-part fix:
1. Pass `--encoding=UTF8 --locale=C` via `initdbFlags` to all
EmbeddedPostgres constructors so the cluster defaults to UTF-8.
2. Explicitly set `encoding 'UTF8'` in the CREATE DATABASE statement
with `template template0` (required because template1 may already
have a different encoding) and `C` locale for portability.
Existing databases created with SQL_ASCII are NOT automatically fixed;
users must delete their local `data/db` directory and restart to
re-initialize the cluster.
Relates to #636
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This shortcut interfered with browser tab-switching (Cmd+1..9) and
produced a black screen when used. Removes the handler, the Layout
callback, and the design-guide documentation entry.
Closes RUS-56
The OpenClaw Gateway's agent method has strict parameter validation
that rejects unknown properties. The paperclip property was being
sent at the root level of agentParams, causing validation failures
with error: "invalid agent params: at root: unexpected property 'paperclip'"
The paperclip metadata is already included in the message field
via wakeText, so removing the separate paperclip property resolves
the validation error while preserving the necessary information.
Fixes#606
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add commit co-author requirement for Paperclip agents and fix markdown
lint issues (blank lines before lists, table column alignment).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Run pnpm install in the new worktree directory before initializing the
Paperclip instance so that node_modules are ready when the init step runs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add global font smoothing (antialiased) to body
- Add tabular-nums to all numeric displays: MetricCard values, Costs page,
AgentDetail token/cost grids and tables, IssueDetail cost summary,
Companies page budget display
- Replace markdown image hard border with subtle inset box-shadow overlay
- Replace all animate-ping status dots with calmer animate-pulse across
AgentDetail, IssueDetail, Agents, sidebar, kanban, issues list, and
active agents panel
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds support for Hermes Agent (https://github.com/NousResearch/hermes-agent)
as a managed employee in Paperclip companies.
Hermes Agent is a full-featured AI agent by Nous Research with 30+ native
tools, persistent memory, session persistence, 80+ skills, MCP support,
and multi-provider model access.
Changes:
- Add 'hermes_local' to AGENT_ADAPTER_TYPES (packages/shared)
- Add @nousresearch/paperclip-adapter-hermes dependency (server)
- Register hermesLocalAdapter in the adapter registry (server)
The adapter package is maintained at:
https://github.com/NousResearch/hermes-paperclip-adapter
Add a "Project" filter section to the issues filter popover, following the
same pattern as the existing Assignee and Labels filters. Issues can now
be filtered by one or more projects from the filter dropdown.
Closes#129
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The paperclip button in comments uploaded images to the issue-level
attachment section but didn't insert a markdown image reference into
the comment body. Now it uses the imageUploadHandler to get the URL
and appends an inline image to the comment text.
Fixes#272
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The codex-local and cursor-local adapters already wrap the
instructionsFilePath read in try/catch, logging a warning and
continuing without instructions. The claude-local adapter was missing
this handling, causing ENOENT crashes when the instructions file
doesn't exist.
Fixes#529
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The server builds its hostname allow-set once at startup. When users
add a new hostname via the CLI, the config file is updated but the
running server doesn't reload it. This adds a clear message telling
users to restart the server for the change to take effect.
Fixes#538
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* 'master' of github.com-dotta:paperclipai/paperclip:
Fix approvals service idempotency test
Add workspace strategy plan doc
Copy git hooks during worktree init
* public-gh/master:
Rebind seeded project workspaces to the current worktree
Copy seeded secrets key into worktree instances
server: make approval retries idempotent (#499)
fix: address review feedback — stale error message and * wildcard
Update server/src/routes/assets.ts
feat: make attachment content types configurable via env var
fix: wire parentId query filter into issues list endpoint
* public-gh/master:
Copy seeded secrets key into worktree instances
server: make approval retries idempotent (#499)
Fix doctor summary after repairs
Fix worktree minimal clone startup
Add minimal worktree seed mode
Add worktree init CLI for isolated development instances
fix: address review feedback — stale error message and * wildcard
Update server/src/routes/assets.ts
feat: make attachment content types configurable via env var
fix: wire parentId query filter into issues list endpoint
Apply suggestions from code review
fix(adapter-utils): strip Claude Code env vars from child processes
- Update stale doc comment in index.ts to reflect direct ~/.gemini/skills/
injection instead of tmpdir approach
- Remove bare GEMINI_API_KEY/GOOGLE_API_KEY from auth regex to prevent
false positives when those strings appear in assistant output
- Align hello probe sandbox/approvalMode flags with execute.ts
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
GEMINI_CLI_HOME pointed to a tmpdir which broke OAuth auth since the CLI
couldn't find credentials in the real home directory.
Instead, inject Paperclip skills directly into ~/.gemini/skills/ (matching
the pattern used by cursor, codex, pi, and opencode adapters). This lets
the Gemini CLI find both auth credentials and skills in their natural
location without any GEMINI_CLI_HOME override.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The Gemini CLI supports OAuth login via `gemini auth login` which stores
credentials locally without setting any env vars. The previous warn-level
check on missing GEMINI_API_KEY caused false alarms when CLI-based OAuth
was used. The hello probe that follows is the real auth authority — if
auth is actually broken, it will catch it and report appropriately.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Incorporate improvements from PR #13 and #105 into the gemini-local adapter:
- Add detectGeminiAuthRequired() for runtime auth failure detection with
errorCode: "gemini_auth_required" on execution results
- Add isGeminiTurnLimitResult() to detect exit code 53 / turn_limit status
and clear session to prevent stuck sessions on next heartbeat
- Add describeGeminiFailure() for structured error messages from parsed
result events including errors array extraction
- Return parsed resultEvent in resultJson instead of raw stdout/stderr
- Add isRetry guard to prevent stale session ID fallback after retry
- Replace boolean yolo with approvalMode string (default/auto_edit/yolo)
with backwards-compatible config.yolo fallback
- Add sandbox config option (--sandbox / --sandbox=none)
- Add GOOGLE_GENAI_USE_GCA auth detection in environment test
- Consolidate auth detection regex into shared detectGeminiAuthRequired()
- Add gemini-2.0-flash and gemini-2.0-flash-lite model IDs
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Isolate skills injection using a temporary directory mapped via
GEMINI_CLI_HOME, mirroring the claude-local sandbox approach
instead of polluting the global ~/.gemini/skills directory.
- Update the environment probe to use `--output-format stream-json`
so the payload matches the downstream parseGeminiJsonl parser.
- Deduplicate `firstNonEmptyLine` helper by extracting it to a
shared `utils.ts` module.
- Clean up orphaned internal exports and update adapter documentation.
Add PAPERCLIP_ALLOWED_ATTACHMENT_TYPES env var to configure allowed
MIME types for issue attachments and asset uploads. Supports exact
types (application/pdf) and wildcard patterns (image/*, text/*).
Falls back to the existing image-only defaults when the env var is
unset, preserving backward compatibility.
- Extract shared module `attachment-types.ts` with `isAllowedContentType()`
and `matchesContentType()` (pure, testable)
- Update `routes/issues.ts` and `routes/assets.ts` to use shared module
- Add unit tests for parsing and wildcard matching
Closes#487
The parentId parameter on GET /api/companies/:companyId/issues was
silently ignored — the filter was never extracted from the query string,
never passed to the service layer, and the IssueFilters type did not
include it. All other filters (status, assigneeAgentId, projectId, etc.)
worked correctly.
This caused subtask lookups to return every issue in the company instead
of only children of the specified parent.
Changes:
- Add parentId to IssueFilters interface
- Add eq(issues.parentId, filters.parentId) condition in list()
- Extract parentId from req.query in the route handler
Fixes: LAS-101
When the Paperclip server is started from within a Claude Code session
(e.g. `npx paperclipai run` in a Claude Code terminal), the `CLAUDECODE`
and related env vars (`CLAUDE_CODE_ENTRYPOINT`, `CLAUDE_CODE_SESSION`,
`CLAUDE_CODE_PARENT_SESSION`) leak into `process.env`. Since
`runChildProcess()` spreads `process.env` into the child environment,
every spawned `claude` CLI process inherits these vars and immediately
exits with: "Claude Code cannot be launched inside another Claude Code
session."
This is particularly disruptive for the `claude-local` adapter, where
every agent run spawns a `claude` child process. A single contaminated
server start (or cron job that inherits the env) silently breaks all
agent executions until the server is restarted in a clean environment.
The fix deletes the four known Claude Code nesting-guard env vars from
the merged environment before passing it to `spawn()`.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
After onboarding, the wizard navigated to the newly created issue
(e.g. /JAR/issues/JAR-1). useCompanyPageMemory then saved this path,
causing every subsequent company switch to land on that stale issue
instead of the dashboard.
Remove the issue-specific navigation branch so handleLaunch always
falls through to the dashboard route.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Scaffolds end-to-end testing with Playwright for the onboarding wizard.
Runs in skip_llm mode by default (UI-only, no LLM costs). Set
PAPERCLIP_E2E_SKIP_LLM=false for full heartbeat verification.
- tests/e2e/playwright.config.ts: Playwright config with webServer
- tests/e2e/onboarding.spec.ts: 4-step wizard flow test
- .github/workflows/e2e.yml: manual workflow_dispatch CI workflow
- package.json: test:e2e and test:e2e:headed scripts
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
All auth config literals in the CLI were missing the required
disableSignUp field after it was added to authConfigSchema.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Simplify the PR-based flow: force push to update the branch if it
already exists, and only create a new PR when one doesn't exist yet.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Use lockfile-bot name/email instead of github-actions[bot]
- Remove force push: close any stale PR and delete branch first,
then create a fresh branch and PR each time
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* public-gh/master:
fix: disable secure cookies for HTTP deployments
feat(adapters): add claude-sonnet-4-6 and claude-haiku-4-6 models
Add opencode-ai to global npm install in Dockerfile
fix: correct env var priority for authDisableSignUp
Add pi-local package.json to Dockerfile
feat: add auth.disableSignUp config option
refactor: extract roleLabels to shared constants
fix(secrets): add secretKeys tracking to resolveEnvBindings for consistent redaction
fix(db): reuse MIGRATIONS_FOLDER constant instead of recomputing
fix(server): wake agent when issue status changes from backlog
fix(server): use home-based path for run logs instead of cwd
fix(db): use fileURLToPath for Windows-safe migration paths
fix(server): auto-deduplicate agent names on creation instead of rejecting
feat(ui): show human-readable role labels in agent list and properties
fix(ui): prevent blank screen when prompt template is emptied
fix(server): redact secret-sourced env vars in run logs by provenance
fix(cli): split path and query in buildUrl to prevent %3F encoding
fix(scripts): use shell on Windows to fix spawn EINVAL in dev-runner
Replace peter-evans/create-pull-request with plain gh CLI commands to
avoid third-party supply chain risk. Uses only GitHub's own tooling
(GITHUB_TOKEN + gh CLI) to create the lockfile refresh PR.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The refresh-lockfile workflow was pushing directly to master, which fails
with branch protection rules. Convert to use peter-evans/create-pull-request
to create a PR instead. Exempt the bot's branch from the lockfile policy check.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Agents run unattended and cannot respond to interactive permission
prompts from Claude Code. When dangerouslySkipPermissions is false
(the previous default), Claude Code blocks file operations with
"Claude requested permissions to write to /path, but you haven't
granted it yet" — making agents unable to edit files.
The OnboardingWizard already sets this to true for claude_local
agents (OnboardingWizard.tsx:277), but agents created or edited
outside the wizard inherit the default of false, breaking them.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Heartbeat runs recorded costs via direct SQL inserts into costEvents and
agents.spentMonthlyCents, bypassing costService.createEvent(). This skipped:
- companies.spentMonthlyCents update (company budget never incremented)
- Agent auto-pause when budget exceeded (enforcement gap)
Now calls costService(db).createEvent() which handles all three:
insert cost event, update agent spend, update company spend, and
auto-pause agent when budgetMonthlyCents is exceeded.
Fixes login failing silently on authenticated + private deployments
served over plain HTTP (e.g. Tailscale, LAN). Users can sign up and
sign in, but the session cookie is rejected by the browser so they
are immediately redirected back to the login page.
Better Auth defaults to __Secure- prefixed cookies with the Secure
flag when NODE_ENV=production. Browsers silently reject Secure cookies
on non-HTTPS origins. This detects when PAPERCLIP_PUBLIC_URL uses
http:// and sets useSecureCookies: false so session cookies work
without HTTPS.
Switch the production stage to the built-in node user from
node:lts-trixie-slim, fixing two runtime failures:
1. Claude CLI rejects --dangerously-skip-permissions when the
process UID is 0, making the claude-local adapter unusable.
2. The server crashed at startup (EACCES) because /paperclip was
root-owned and the process could not write logs or instance data.
Changes vs the naive fix:
- Use COPY --chown=node:node instead of a separate RUN chown -R,
avoiding a duplicate image layer that would double the size of
the /app tree in the final image.
- Consolidate mkdir /paperclip + chown into the same RUN layer as
the npm global install (already runs as root) to keep layer count
minimal.
- Add USER node before CMD so the process runs unprivileged.
The VOLUME declaration comes after chown so freshly-mounted
anonymous volumes inherit the correct node:node ownership.
Fixes#344
- Move status icon to left column on mobile across issues list, inbox, and dashboard
- Hide priority icon on mobile (only show on desktop)
- Move unread indicator dot to right side vertically centered on mobile inbox
- Stale work section: show status icon instead of clock on mobile
- Desktop layout unchanged
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The description text was being covered by the property chips bar when
typing long content on iOS with the virtual keyboard open.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Env var now properly overrides file config in both directions
- Follows established pattern for boolean config flags
- Removed redundant ?? false (field is typed boolean)
- PAPERCLIP_AUTH_DISABLE_SIGN_UP can now set to 'false' to
override file config's 'true'
Replace <button> with <span role="button"> for the unread dot to
eliminate browser UA button styles (min-height, padding) that caused
unread rows to be taller despite explicit h-4 constraint. Also
reduce dot from h-2.5/w-2.5 to h-2/w-2 for a more subtle indicator.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Removed the scroll-to-bottom button from IssuesList (wrong location)
and created a shared ScrollToBottom component. Added it to IssueDetail
and RunDetail pages. On mobile, the button sits above the bottom nav
to avoid overlap.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Moves the unread dot from a separate left column into the metadata
line (alongside status/identifier), with an empty placeholder for
read items to keep spacing consistent. Reduces left padding on
mobile inbox rows to reclaim space.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add PriorityIcon and timeAgo to IssuesList mobile rows to match the
pattern used in Inbox and Dashboard. Align Dashboard row padding to
match Inbox. All mobile issue rows now show: title (2-line clamp),
then priority + status + identifier + relative time.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
On mobile, all issue rows now show title first (up to 2 lines),
with metadata (icons, identifier, timestamp) on a second line below.
Desktop layout is preserved as single-line rows.
Affected locations: Inbox recent issues, Inbox stale work,
Dashboard recent tasks, and IssuesList.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Give unread dot container fixed h-5 so rows are consistent height
regardless of badge presence
- Use flex-wrap on mobile so title gets its own line with line-clamp-2
- On sm+ screens, keep single-line truncated layout
- Move timestamp to ml-auto with sm:order-last for clean wrapping
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add min-w-0 and overflow-hidden to the stale work row flex containers
so the title truncates properly on narrow screens. Add shrink-0 to
identifier and assignee spans to prevent them from being compressed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Added disableSignUp to authConfigSchema in config-schema.ts
- Added authDisableSignUp to Config interface
- Added parsing from PAPERCLIP_AUTH_DISABLE_SIGN_UP env or config file
- Passed to better-auth emailAndPassword.disableSignUp
When true, blocks new user registrations on public instances.
Defaults to false (backward compatible).
Fixes#241
Move duplicated roleLabels map from AgentProperties.tsx, Agents.tsx,
OrgChart.tsx, and agent-config-primitives.tsx into AGENT_ROLE_LABELS
in packages/shared/src/constants.ts.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
resolveEnvBindings now returns { env, secretKeys } matching the pattern
already used by resolveAdapterConfigForRuntime, so any caller can redact
secret-sourced values by provenance rather than key-name heuristics alone.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The local migrationsFolder variable in migratePostgresIfEmpty duplicated
the module-level MIGRATIONS_FOLDER constant. Reuse the constant to keep
a single source of truth for the migration path.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* public-gh/master:
Revert lockfile changes from OpenClaw gateway cleanup
Enhance plugin architecture by introducing agent tool contributions and plugin-to-plugin communication. Update workspace plugin definitions for direct OS access and refine UI extension surfaces. Document new capabilities for plugin settings UI and lifecycle management.
Remove legacy OpenClaw adapter and keep gateway-only flow
Fix CI typecheck and default OpenClaw sessions to issue scope
fix(server): serve cached index.html in SPA catch-all to prevent 500
Add CEO OpenClaw invite endpoint and update onboarding UX
openclaw gateway: auto-approve first pairing and retry
openclaw gateway: persist device keys on create/update and clarify pairing flow
plugin spec draft ideas v0
openclaw gateway: persist device keys and smoke pairing flow
openclaw-gateway: document and surface pairing-mode requirements
fix(smoke): disambiguate case C ack comment target
fix(openclaw-gateway): enforce join token validation and add smoke preflight gates
fix(openclaw): make invite snippet/onboarding gateway-first
res.sendFile can emit NotFoundError from the send module in certain
path resolution scenarios, causing 500s on company-scoped SPA routes.
Cache index.html at startup and serve it directly, which is both
more reliable and faster.
Fixes#233
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously, agents were only woken when the assignee changed. Now
also wakes the assigned agent when an issue transitions out of
backlog status (e.g. backlog -> todo).
Fixes#167
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Run logs defaulted to process.cwd()/data/run-logs, placing logs in
unexpected locations when launched from non-home directories. Now
defaults to ~/.paperclip/instances/<id>/data/run-logs/.
Fixes#89
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
URL.pathname returns /C:/... on Windows, causing doubled drive letters
when Node prepends the current drive. fileURLToPath handles this
correctly across platforms.
Fixes#132
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace assertCompanyShortnameAvailable with deduplicateAgentName in
the create path so duplicate names get auto-suffixed (e.g. Engineer 2)
instead of throwing a conflict error.
Fixes#232
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use roleLabels lookup in list view subtitle and AgentProperties
panel instead of raw role strings.
Fixes#180
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Change onChange handler from v || undefined to v ?? "" so empty
strings don't become undefined and crash downstream .trim() calls.
Fixes#191
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
resolveAdapterConfigForRuntime now returns a secretKeys set tracking
which env vars came from secret_ref bindings. The onAdapterMeta
callback uses this to redact them regardless of key name.
Fixes#234
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The URL constructor's pathname setter encodes ? as %3F, breaking
heartbeat event polling. Split query params before assignment.
Fixes#204
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add 400 response for /issues without companyId, tag issue.updated
activity with source:comment when triggered by a comment, and mark
comment activities with updated:true when field changes accompany them.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extract attachErrorContext helper to DRY up the error handler, attach the
original Error object to res.err so pino can serialize stack traces, and
rename the log context key from err to errorContext so it doesn't clash
with pino's built-in err serializer.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add catch-all handler after API router to return a proper 404 JSON
response instead of falling through to the SPA handler.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Update cwd test to expect an error for missing directories (matches
createIfMissing: false accepted from review)
- Add warn-level check for non-ProviderModelNotFoundError failures
during best-effort model discovery when no model is configured
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Move costUsd to top-level return field in parseOpenCodeJsonl (out of usage)
- Fix session-not-found regex to match "Session not found" pattern
- Use callID for toolUseId in UI stdout parser, add status/metadata header
- Fix CLI formatter: separate tool_call/tool_result lines, split step_finish
- Enable createIfMissing for cwd validation in environment tests
- Add empty OPENAI_API_KEY override detection
- Classify ProviderModelNotFoundError as warning during model discovery
- Make model discovery best-effort when no model is configured
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Update cwd test to expect an error for missing directories (matches
createIfMissing: false accepted from review)
- Add warn-level check for non-ProviderModelNotFoundError failures
during best-effort model discovery when no model is configured
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Move costUsd to top-level return field in parseOpenCodeJsonl (out of usage)
- Fix session-not-found regex to match "Session not found" pattern
- Use callID for toolUseId in UI stdout parser, add status/metadata header
- Fix CLI formatter: separate tool_call/tool_result lines, split step_finish
- Enable createIfMissing for cwd validation in environment tests
- Add empty OPENAI_API_KEY override detection
- Classify ProviderModelNotFoundError as warning during model discovery
- Make model discovery best-effort when no model is configured
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a "+" button next to "AGENTS" in the sidebar, matching the existing
pattern used by Projects. Clicking it opens the agent creation choice
modal.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Short titles now always occupy two lines of vertical space, ensuring
consistent horizontal alignment across agent cards.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds a small copy icon to the right of each comment's date that copies
the comment body as raw markdown to the clipboard. Shows a checkmark
briefly after copying.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Document observed failures before wake-text fix and successful
stock-clean lane pass after fix. Note instrumentation lane limitations.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Include the full Paperclip API workflow (checkout, fetch, update) and
endpoint bans in the wake text sent to OpenClaw agents, preventing
them from guessing non-existent endpoints. Applied to both openclaw
and openclaw_gateway adapters.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Major improvements to the Pi local adapter:
Bug Fixes (Greptile-identified):
- Fix string interpolation in models.ts error message (was showing literal ${detail})
- Fix tool matching in parse.ts to use toolCallId instead of toolName for correct
multi-call handling and result assignment
- Fix dead code in execute.ts by tracking instructionsReadFailed flag
Feature Improvements:
- Switch from print mode (-p) to RPC mode (--mode rpc) to prevent agent from
exiting prematurely and ensure proper lifecycle completion
- Add stdin command sending via JSON-RPC format for prompt delivery
- Add line buffering in execute.ts to handle partial JSON chunks correctly
- Filter RPC protocol messages (response, extension_ui_request, etc.) from transcript
Cost Tracking:
- Extract cost and usage data from turn_end assistant messages
- Support both Pi format (input/output/cacheRead/cost.total) and generic format
- Add tests for cost extraction and accumulation across multiple turns
All tests pass (12/12), typecheck clean, server builds successfully.
Capture run events, logs, issue state, and container logs on failures
or timeouts for debugging. Write compatibility JSON keys for claimed
API key. Add two-lane validation requirement to test plan.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Document the agent local-cli command for manual CLI usage and add a
step-by-step self-test playbook for validating assignment flows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New subcommand to install Paperclip skills for Claude/Codex agents and
print the required PAPERCLIP_* environment variables for local CLI
usage outside heartbeat runs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Lazy-load mermaid.js and render fenced mermaid code blocks as inline
SVG diagrams with dark/light mode support. Falls back to showing the
source code on parse errors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Ensure unique URL-safe shortnames by appending numeric suffixes when
collisions occur. Applied during project creation, update, and company
import flows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New adapter type for invoking OpenClaw agents via the gateway protocol.
Registers in server, CLI, and UI adapter registries. Adds onboarding
wizard support with gateway URL field and e2e smoke test script.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace the direct agent config dialog with a choice modal offering two
paths: "Ask the CEO to create a new agent" (opens pre-filled issue) or
"I want advanced configuration myself" (navigates to /agents/new).
- Extend NewIssueDefaults with title/description for pre-fill support
- Add /agents/new route with full-page agent configuration form
- NewAgentDialog now shows CEO recommendation modal
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When a company has "require board approval for new agents" enabled,
hiring an agent creates a pending approval that requires the user
(as a board member) to approve before the agent can start working.
However, the sidebar inbox badge did not include pending approvals
in its count, so there was no visual indicator that action was needed.
Users had no way of knowing an approval was waiting unless they
happened to open the Inbox page manually.
The root cause: the sidebar-badges service correctly included
approvals in the inbox total, but the route handler overwrites
badges.inbox to add alertsCount and staleIssueCount — and in
doing so dropped badges.approvals from the sum.
Add badges.approvals to the inbox count recalculation so that
pending and revision-requested approvals surface in the sidebar
notification badge alongside failed runs, alerts, stale work,
and join requests.
Affected files:
- server/src/routes/sidebar-badges.ts
Change the issue title in agent run cards from single-line truncate
to line-clamp-2 so titles always occupy two lines for consistent card height.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add create_github_release helper to release script that creates a
GitHub Release via gh CLI after tagging, using release notes from
releases/vX.Y.Z.md if available or auto-generated notes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Accept x-openclaw-token as the preferred auth header for OpenClaw
invite/join flows, falling back to x-openclaw-auth for backwards
compatibility. Update diagnostics messages accordingly.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add OpenClawEndpointKind type to distinguish between /hooks/wake,
/hooks/agent, open_responses, and generic endpoints. Build appropriate
payloads per endpoint kind with optional sessionKey inclusion.
Refactor webhook execution to use endpoint-aware payload construction.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Introduced SVG sanitization using `dompurify` to prevent malicious content.
- Updated tests to validate SVG sanitization with various scenarios.
- Enhanced response headers for assets, adding CSP and nosniff for SVGs.
- Adjusted UI to better clarify supported file types for logo uploads.
- Updated dependencies to include `jsdom` and `dompurify`.
Agents were being woken immediately when an issue was created or
reassigned with status "backlog", defeating the purpose of backlog
as "not ready to work on yet" (issue #96).
Added `&& issue.status !== "backlog"` guard to both the CREATE and
PATCH wakeup paths, mirroring the existing done/cancelled suppression
pattern already present in the file.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Webhook auth header and gateway auth token fields now render as password
inputs by default, with an eye/eye-off toggle icon on the left to reveal
the value.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When a join request arrives via WebSocket live event, display a toast
prompting the user to review it in the inbox.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
In the My Recent Issues section, the blue unread dot is now a button that
marks the issue as read on click with a smooth opacity fade-out. Already-read
issues show empty space instead of a hollow circle.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add skill URL note to onboarding text document and strengthen callback
reachability test language in company settings invite snippet.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Merge consecutive assistant/thinking deltas into a single feed entry
instead of creating one per chunk. Add dedupeKey to FeedItem, increase
streaming text cap to 4000 chars, and bump seen-keys limit to 6000.
Applied consistently to both ActiveAgentsPanel and LiveRunWidget.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* public-gh/master:
fix(auth): apply effective trusted origins and honor allowed hostnames in public mode
fix(onboard): preserve env-derived secrets defaults and report ignored exposure env in local_trusted mode
fix: parseBooleanFromEnv silently treats common truthy values as false
set `PAPERCLIP_PUBLIC_URL` in compose files
centralize URLs into single canonical URL var
When approving an agent join request with a shortname already in use,
append a numeric suffix (e.g. "openclaw-2") instead of returning an error.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The dark-mode --destructive color was too dim (lightness 0.396) to read
against dark backgrounds. Bumped to 0.637 lightness with higher chroma
so error text is clearly visible. Set --destructive-foreground to white
for contrast on destructive button backgrounds.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The sidebar badge count was missing join requests from its inbox total,
and the live updates provider had no handler for join_request entity
type, so new join requests wouldn't appear until manual page refresh.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The previous fix (8151331) set res.err but pino-http wasn't picking it
up (likely Express 5 response object behavior). Switch to a custom
__errorContext property on the response that customErrorMessage and
customProps read directly, bypassing pino-http's unreliable res.err
check. Remove duplicate manual logger.error calls from the error
handler since pino-http now gets the full context.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The server-side badge counted agent error alerts independently of failed
runs, but the UI suppresses agent error alerts when individual failed run
cards are already shown. This mismatch caused the badge to show e.g. 2
while only 1 item was visible. Align server logic with the client.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- pino-http customErrorMessage now includes the real error message
- customProps includes reqBody, reqParams, reqQuery, and routePath for 4xx/5xx
- Error handler logs full request context (body, params, query) for both
HttpError 500s and unhandled errors
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove duplicate opencode_local adapter entry in OnboardingWizard
(old Code-icon version), keeping only the OpenCodeLogoIcon entry
- Extract resolveOpenCodeCommand() helper to deduplicate the
PAPERCLIP_OPENCODE_COMMAND env-var fallback logic in models.ts
- Bump @types/node from ^22.12.0 to ^24.6.0 to match the monorepo
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* public-gh/master:
fix(ui): wrap failed run card actions on mobile
feat(codex): add gpt-5.4 to codex_local model list
persist paperclip data in a named volume
add support to `cursor` and `opencode` in containerized instances
force `@types/node@24` in the server
Add artifact-check to fail fast on broken builds
remove an insecure default auth secret
expose `PAPERCLIP_ALLOWED_HOSTNAMES` in compose files
wait for a health db
update lock file
update typing to node v24 from v20
add missing `openclaw` adapter from deps stage
fix incorrect pkg scope
update docker base image
move docker into `authenticated` deployment mode
When "Generate agent snippet" is clicked, the snippet is now
automatically copied to the clipboard and the "Copied" delight
animation is shown, matching the existing manual copy behavior.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Same issue as the UI fix — merge conflict resolution kept HEAD's
reference to the removed constant. OpenCode uses strict model
selection with no default.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
PR #62 uses strict model selection with no default — the merge
conflict resolution incorrectly kept HEAD's references to this
removed constant. Also remove dead opencode_local branch in
OnboardingWizard (already handled by prior condition).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The lockfile was out of sync with the merged package.json files
(adapter-utils @types/node bumped to ^24.6.0 by PR #62).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Merges paperclipai/paperclip#62 onto latest master (494448d).
Adds complete OpenCode provider with strict model selection,
dynamic model discovery, CLI/server/UI adapter registration.
Resolved conflicts with master's cursor adapter additions,
node v24 typing, and containerized opencode support (201d91b).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The cache-first strategy for static assets was causing pages to serve
stale content on refresh. Switched to network-first for all requests
with cache as offline-only fallback. Bumped cache version to clear
existing caches on activate.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
setSearchParams from React Router triggers router state updates that cause
the Issues component tree to re-render on each debounced URL update. Switch
to window.history.replaceState which updates the URL silently without
triggering React Router re-renders.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Radix Dialog wraps content in react-remove-scroll, which blocks wheel
events on portaled Popover content (rendered outside the Dialog DOM
tree). Add a disablePortal option to PopoverContent and use it for all
InlineEntitySelector instances inside NewIssueDialog so the Popover
stays in the Dialog's DOM tree and scrolling works.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add renderTriggerValue/renderOption to the comment thread's assignee
selector so it shows agent icons, matching the new issue dialog. Fix
the InlineEntitySelector flash on click by only auto-opening on
keyboard focus (not pointer-triggered focus).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Debounces search input (300ms) and syncs it to a ?q= URL parameter so
searches persist across navigation and can be shared via URL.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace scrollbar-none with a scrollbar-auto-hide utility that keeps the
scrollbar thumb transparent by default and reveals it on container hover.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Failed runs, alerts, and stale work items can now be dismissed via an
X button that appears on hover. Dismissed items are stored in
localStorage and filtered from the inbox view and item count.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When served via npx, the absolute path to index.html traverses .npm,
triggering Express 5 / send's dotfile guard. Using the root option
makes send only check the relative filename for dotfiles.
Fixes#48
Shows a floating arrow button in the bottom-right corner when the user
is more than 300px from the bottom of the issues list. Hides
automatically when near the bottom. Smooth-scrolls on click.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a ProjectStatusPicker component that renders the status badge as a
clickable button opening a popover with all project statuses. Falls back
to the read-only StatusBadge when no onUpdate handler is provided.
Works in both desktop side panel and mobile sheet layout.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
On mobile, the sidebar panel toggle was hidden (md:flex only). Add a
mobile-visible SlidersHorizontal button that opens a bottom Sheet drawer
with ProjectProperties, matching the existing pattern from IssueDetail.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When searching via the command palette on mobile, the left sidebar
now automatically closes so search results are visible.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Force popover to always open downward (side="bottom") to prevent it from
flipping upward and going off-screen on mobile. Skip auto-focusing the
search input on touch devices so the virtual keyboard doesn't open and
reshape the viewport. Add touch-manipulation on option buttons to remove
tap delays and improve scroll gesture recognition.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The X close button in the command palette dialog used the generic dialog
positioning (absolute top-4 right-4), which was visually offset from the
search input on mobile. Replace with a custom close button that matches
the input wrapper height (h-12) and uses flex centering for proper
vertical alignment.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Allow the "For [Assignee] in [Project]" row to wrap on mobile instead of
requiring horizontal scroll that pushes selectors off-screen. On desktop
(sm+), the row stays inline with min-w-max. Added overscroll-x-contain
for better touch scroll containment.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
On small screens, the assignee Identity badge takes up valuable space
in the command palette search results, truncating issue titles. Hide
it below the `sm` breakpoint so more of the title is visible.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Radix Dialog's modal DismissableLayer calls preventDefault() on pointerdown
events originating outside the Dialog DOM tree. Popover portals render at the
body level (outside the Dialog), so touch events on popover content were
treated as 'outside' — killing scroll gesture recognition on mobile.
Fix: add onPointerDownOutside to NewIssueDialog's DialogContent that detects
events from Radix popper wrappers and calls event.preventDefault() on the
Radix event (not the native event), which skips the Dialog's native
preventDefault and restores touch scrolling.
Also cleans up previous CSS-only workarounds (-webkit-overflow-scrolling,
touch-pan-y on individual buttons) that couldn't override JS preventDefault.
Adds service worker with network-first navigation (SPA fallback) and
cache-first static assets. Enhances web manifest with start_url, scope,
id, description, orientation, and maskable icon entry.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add OpenResponses gateway enablement instructions to the end of the
agent snippet in CompanySettings. Refactor buildAgentSnippet to use
a template literal for easier future editing.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Support two-phase canary release flow:
- --canary: publishes packages under @canary npm tag, skips git commit/tag
- --promote <version>: moves canary packages to @latest, then commits and tags
Both flags work with --dry-run. Existing behavior unchanged when neither flag is passed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New Steps 3-5 in the release skill:
- Step 3: Publish canary to npm with --tag canary (latest untouched)
- Step 4: Smoke test canary in Docker (uses existing smoke infrastructure)
- Step 5: Promote canary to latest after verification
Also updates idempotency table for new intermediate states (canary
published but not promoted) and adds release flow summary.
Script changes still needed: --canary flag for release.sh, --promote
command for dist-tag promotion.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When the properties pane was closed on a project page, there was no way
to reopen it. Add the same SlidersHorizontal toggle button used on issue
detail pages.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The append method created a new WriteStream for every log chunk and resolved
the promise on the end callback (data flushed) rather than the close event
(fd released). Over many agent runs the leaked fds corrupted the fd table,
causing child_process.spawn to fail with EBADF.
Replace with fs.appendFile which properly opens, writes, and closes the fd
before resolving.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
pino-http checks res.err before falling back to its generic
"failed with status code 500" error. Set res.err to the real error
in the error handler so logs include the actual message and stack trace.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Invite link is now automatically copied to clipboard when created
- "Copied" badge appears next to the share link header for 2s
- Standalone "Copy link" button replaced with a small copy icon next to the link
- Icon toggles to a green checkmark while "copied" feedback is shown
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- release-changelog: Step 0 checks for existing changelog file before
generating. Asks reviewer to keep/regenerate/update. Never overwrites
silently. Clarifies this skill never triggers version bumps.
- release: Step 0 idempotency table covering all steps. Tag check before
npm publish prevents double-release. Task search before creation prevents
duplicate follow-up tasks. Supports iterating on changelogs pre-publish.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduces the release-changelog skill (skills/release-changelog/SKILL.md)
that teaches agents to generate user-facing changelogs from git history,
changesets, and merged PRs. Includes breaking change detection, categorization
heuristics, and structured markdown output template.
Also creates the releases/ directory convention for storing versioned
release notes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tracks most recently selected assignees in localStorage and sorts both
the issue properties and new issue dialog assignee lists accordingly.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* 'master' of github.com-dotta:paperclipai/paperclip:
fix(ui): render sub-goals in goal detail tree
fix: exclude terminated agents from list and org chart endpoints
- Comment dates are now clickable anchor links (#comment-{id})
- Pages scroll to and highlight the target comment when URL has a hash
- Added GET /api/issues/:id/comments/:commentId endpoint
- Updated skill docs with new endpoint and comment URL format
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Show only the command for shellToolCall/shell inputs instead of the
full payload. Format shell results with exit code + truncated
stdout/stderr sections for readability.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Agents in pending_approval status are not invokable, but tickTimers only
skipped paused and terminated agents, causing enqueueWakeup to throw a
409 conflict error on every heartbeat tick for pending_approval agents.
Added pending_approval to the skip guard in tickTimers to match the
existing guard in enqueueWakeup.
The --yolo flag bypasses interactive prompts more broadly than --trust.
Updated execute, test probe, docs, and test expectations.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Product spec for ClipHub — a marketplace for Paperclip team
configurations, agent blueprints, skills, and governance templates.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Detect JSON-like tool output and format it with indentation instead of
displaying it as a single compressed line.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Export ensureCursorSkillsInjected from the server entrypoint and add
a test for skill directory injection. Document the auto-inject and
auto-trust behaviors in the adapter notes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The retry button was only on the Inbox page but missing from the individual
run detail view. This adds it next to the status badge, matching the same
wakeup pattern used in the Inbox (retry_failed_run with context snapshot).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
pino-http was logging 500s at INFO level with a generic "failed with status
code 500" message. Now 500s log at ERROR level and include the actual error
(message, stack, name) via res.locals handoff from the error handler.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Update comment style guide and API reference to use /<prefix>/issues/...
format instead of /issues/... so agent-generated markdown links navigate
directly to the correct company-scoped route.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously, sorting was skipped when a search query was active, so
search results were only ordered by backend relevance ranking.
Now client-side sorting applies to search results too.
Also changed default sort from "created" to "updated" desc so most
recently updated issues appear first.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Shows a green pill badge on the Claude Code and Codex adapter cards
during CEO agent configuration in the onboarding flow.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Terminated agents (e.g. from rejected hire approvals) were visible in
GET /companies/:companyId/agents and GET /companies/:companyId/org because
list() and orgForCompany() had no status filtering.
- Add ne(agents.status, "terminated") filter to both queries
- Add optional { includeTerminated: true } param to list() for callers
that need all agents (e.g. company-portability export with skip counting)
- orgForCompany() always excludes terminated (no escape hatch needed)
Fixes#5
Companies with status "archived" are now hidden from the company
dropdown in the NewIssueDialog, matching the expected behavior.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The API already supports full-text search via ?q= on the issues list
endpoint. Added documentation to both SKILL.md and the API reference
so agents know they can search issues by title, identifier,
description, and comments.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Instead of showing alarming warnings on first run when storage, log,
and database directories don't exist, the doctor checks now silently
create them and report pass. LLM provider is treated as optional
rather than a warning when not configured.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove Docker option from quickstart, make `npx paperclipai onboard --yes` the recommended path
- Add tip about `create-agent-adapter` skill in the creating-an-adapter guide
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The release script cleaned up temporary build artifacts (ui-dist, skills)
before staging files for the commit, then tried to git add server/ui-dist
which no longer existed. Remove it from the git add list since these
artifacts should never be committed — they're only needed during publish.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-03 16:14:28 -06:00
936 changed files with 325790 additions and 8908 deletions
Create agent company packages conforming to the Agent Companies specification
(agentcompanies/v1). Use when a user wants to create a new agent company from
scratch, build a company around an existing git repo or skills collection, or
scaffold a team/department of agents. Triggers on: "create a company", "make me
a company", "build a company from this repo", "set up an agent company",
"create a team of agents", "hire some agents", or when given a repo URL and
asked to turn it into a company. Do NOT use for importing an existing company
package (use the CLI import command instead) or for modifying a company that
is already running in Paperclip.
---
# Company Creator
Create agent company packages that conform to the Agent Companies specification.
Spec references:
- Normative spec: `docs/companies/companies-spec.md` (read this before generating files)
- Web spec: https://agentcompanies.io/specification
- Protocol site: https://agentcompanies.io/
## Two Modes
### Mode 1: Company From Scratch
The user describes what they want. Interview them to flesh out the vision, then generate the package.
### Mode 2: Company From a Repo
The user provides a git repo URL, local path, or tweet. Analyze the repo, then create a company that wraps it.
See [references/from-repo-guide.md](references/from-repo-guide.md) for detailed repo analysis steps.
## Process
### Step 1: Gather Context
Determine which mode applies:
- **From scratch**: What kind of company or team? What domain? What should the agents do?
- **From repo**: Clone/read the repo. Scan for existing skills, agent configs, README, source structure.
### Step 2: Interview (Use AskUserQuestion)
Do not skip this step. Use AskUserQuestion to align with the user before writing any files.
**For from-scratch companies**, ask about:
- Company purpose and domain (1-2 sentences is fine)
- What agents they need - propose a hiring plan based on what they described
- Whether this is a full company (needs a CEO) or a team/department (no CEO required)
- Any specific skills the agents should have
- How work flows through the organization (see "Workflow" below)
- Whether they want projects and starter tasks
**For from-repo companies**, present your analysis and ask:
- Confirm the agents you plan to create and their roles
- Whether to reference or vendor any discovered skills (default: reference)
- Any additional agents or skills beyond what the repo provides
- Company name and any customization
- Confirm the workflow you inferred from the repo (see "Workflow" below)
**Workflow — how does work move through this company?**
A company is not just a list of agents with skills. It's an organization that takes ideas and turns them into work products. You need to understand the workflow so each agent knows:
- Who gives them work and in what form (a task, a branch, a question, a review request)
- What they do with it
- Who they hand off to when they're done, and what that handoff looks like
- What "done" means for their role
**Not every company is a pipeline.** Infer the right workflow pattern from context:
- **Pipeline** — sequential stages, each agent hands off to the next. Use when the repo/domain has a clear linear process (e.g. plan → build → review → ship → QA, or content ideation → draft → edit → publish).
- **Hub-and-spoke** — a manager delegates to specialists who report back independently. Use when agents do different kinds of work that don't feed into each other (e.g. a CEO who dispatches to a researcher, a marketer, and an analyst).
- **Collaborative** — agents work together on the same things as peers. Use for small teams where everyone contributes to the same output (e.g. a design studio, a brainstorming team).
- **On-demand** — agents are summoned as needed with no fixed flow. Use when agents are more like a toolbox of specialists the user calls directly.
For from-scratch companies, propose a workflow pattern based on what they described and ask if it fits.
For from-repo companies, infer the pattern from the repo's structure. If skills have a clear sequential dependency (like `plan-ceo-review → plan-eng-review → review → ship → qa`), that's a pipeline. If skills are independent capabilities, it's more likely hub-and-spoke or on-demand. State your inference in the interview so the user can confirm or adjust.
**Key interviewing principles:**
- Propose a concrete hiring plan. Don't ask open-ended "what agents do you want?" - suggest specific agents based on context and let the user adjust.
- Keep it lean. Most users are new to agent companies. A few agents (3-5) is typical for a startup. Don't suggest 10+ agents unless the scope demands it.
- From-scratch companies should start with a CEO who manages everyone. Teams/departments don't need one.
- Ask 2-3 focused questions per round, not 10.
### Step 3: Read the Spec
Before generating any files, read the normative spec:
```
docs/companies/companies-spec.md
```
Also read the quick reference: [references/companies-spec.md](references/companies-spec.md)
And the example: [references/example-company.md](references/example-company.md)
### Step 4: Generate the Package
Create the directory structure and all files. Follow the spec's conventions exactly.
**Directory structure:**
```
<company-slug>/
├── COMPANY.md
├── agents/
│ └── <slug>/AGENTS.md
├── teams/
│ └── <slug>/TEAM.md (if teams are needed)
├── projects/
│ └── <slug>/PROJECT.md (if projects are needed)
├── tasks/
│ └── <slug>/TASK.md (if tasks are needed)
├── skills/
│ └── <slug>/SKILL.md (if custom skills are needed)
└── .paperclip.yaml (Paperclip vendor extension)
```
**Rules:**
- Slugs must be URL-safe, lowercase, hyphenated
- COMPANY.md gets `schema: agentcompanies/v1` - other files inherit it
- Agent instructions go in the AGENTS.md body, not in .paperclip.yaml
- Skills referenced by shortname in AGENTS.md resolve to `skills/<shortname>/SKILL.md`
- For external skills, use `sources` with `usage: referenced` (see spec section 12)
- Do not export secrets, machine-local paths, or database IDs
- Omit empty/default fields
- For companies generated from a repo, add a references footer at the bottom of COMPANY.md body:
`Generated from [repo-name](repo-url) with the company-creator skill from [Paperclip](https://github.com/paperclipai/paperclip)`
**Reporting structure:**
- Every agent except the CEO should have `reportsTo` set to their manager's slug
- The CEO has `reportsTo: null`
- For teams without a CEO, the top-level agent has `reportsTo: null`
**Writing workflow-aware agent instructions:**
Each AGENTS.md body should include not just what the agent does, but how they fit into the organization's workflow. Include:
1.**Where work comes from** — "You receive feature ideas from the user" or "You pick up tasks assigned to you by the CTO"
2.**What you produce** — "You produce a technical plan with architecture diagrams" or "You produce a reviewed, approved branch ready for shipping"
3.**Who you hand off to** — "When your plan is locked, hand off to the Staff Engineer for implementation" or "When review passes, hand off to the Release Engineer to ship"
4.**What triggers you** — "You are activated when a new feature idea needs product-level thinking" or "You are activated when a branch is ready for pre-landing review"
This turns a collection of agents into an organization that actually works together. Without workflow context, agents operate in isolation — they do their job but don't know what happens before or after them.
### Step 5: Confirm Output Location
Ask the user where to write the package. Common options:
- A subdirectory in the current repo
- A new directory the user specifies
- The current directory (if it's empty or they confirm)
### Step 6: Write README.md and LICENSE
**README.md** — every company package gets a README. It should be a nice, readable introduction that someone browsing GitHub would appreciate. Include:
- Company name and what it does
- The workflow / how the company operates
- Org chart as a markdown list or table showing agents, titles, reporting structure, and skills
- Brief description of each agent's role
- Citations and references: link to the source repo (if from-repo), link to the Agent Companies spec (https://agentcompanies.io/specification), and link to Paperclip (https://github.com/paperclipai/paperclip)
- A "Getting Started" section explaining how to import: `paperclipai company import --from <path>`
**LICENSE** — include a LICENSE file. The copyright holder is the user creating the company, not the upstream repo author (they made the skills, the user is making the company). Use the same license type as the source repo (if from-repo) or ask the user (if from-scratch). Default to MIT if unclear.
### Step 7: Write Files and Summarize
Write all files, then give a brief summary:
- Company name and what it does
- Agent roster with roles and reporting structure
- Skills (custom + referenced)
- Projects and tasks if any
- The output path
## .paperclip.yaml Guidelines
The `.paperclip.yaml` file is the Paperclip vendor extension. It configures adapters and env inputs per agent.
### Adapter Rules
**Do not specify an adapter unless the repo or user context warrants it.** If you don't know what adapter the user wants, omit the adapter block entirely — Paperclip will use its default. Specifying an unknown adapter type causes an import error.
Paperclip's supported adapter types (these are the ONLY valid values):
-`claude_local` — Claude Code CLI
-`codex_local` — Codex CLI
-`opencode_local` — OpenCode CLI
-`pi_local` — Pi CLI
-`cursor` — Cursor
-`gemini_local` — Gemini CLI
-`openclaw_gateway` — OpenClaw gateway
Only set an adapter when:
- The repo or its skills clearly target a specific runtime (e.g. gstack is built for Claude Code, so `claude_local` is appropriate)
- The user explicitly requests a specific adapter
- The agent's role requires a specific runtime capability
### Env Inputs Rules
**Do not add boilerplate env variables.** Only add env inputs that the agent actually needs based on its skills or role:
-`GH_TOKEN` for agents that push code, create PRs, or interact with GitHub
- API keys only when a skill explicitly requires them
- Never set `ANTHROPIC_API_KEY` as a default empty env variable — the runtime handles this
Example with adapter (only when warranted):
```yaml
schema:paperclip/v1
agents:
release-engineer:
adapter:
type:claude_local
config:
model:claude-sonnet-4-6
inputs:
env:
GH_TOKEN:
kind:secret
requirement:optional
```
Example — only agents with actual overrides appear:
```yaml
schema:paperclip/v1
agents:
release-engineer:
inputs:
env:
GH_TOKEN:
kind:secret
requirement:optional
```
In this example, only `release-engineer` appears because it needs `GH_TOKEN`. The other agents (ceo, cto, etc.) have no overrides, so they are omitted entirely from `.paperclip.yaml`.
## External Skill References
When referencing skills from a GitHub repo, always use the references pattern:
```yaml
metadata:
sources:
- kind:github-file
repo:owner/repo
path:path/to/SKILL.md
commit:<full SHA from git ls-remote or the repo>
attribution:Owner or Org Name
license:<from the repo's LICENSE>
usage:referenced
```
Get the commit SHA with:
```bash
git ls-remote https://github.com/owner/repo HEAD
```
Do NOT copy external skill content into the package unless the user explicitly asks.
Read the local spec file before generating any package files. The spec defines the canonical format and all frontmatter fields. Below is a quick-reference summary for common authoring tasks.
A minimal but complete example of an agent company package.
## Directory Structure
```
lean-dev-shop/
├── COMPANY.md
├── agents/
│ ├── ceo/AGENTS.md
│ ├── cto/AGENTS.md
│ └── engineer/AGENTS.md
├── teams/
│ └── engineering/TEAM.md
├── projects/
│ └── q2-launch/
│ ├── PROJECT.md
│ └── tasks/
│ └── monday-review/TASK.md
├── tasks/
│ └── weekly-standup/TASK.md
├── skills/
│ └── code-review/SKILL.md
└── .paperclip.yaml
```
## COMPANY.md
```markdown
---
name: Lean Dev Shop
description: Small engineering-focused AI company that builds and ships software products
slug: lean-dev-shop
schema: agentcompanies/v1
version: 1.0.0
license: MIT
authors:
- name: Example Org
goals:
- Build and ship software products
- Maintain high code quality
---
Lean Dev Shop is a small, focused engineering company. The CEO oversees strategy and coordinates work. The CTO leads the engineering team. Engineers build and ship code.
```
## agents/ceo/AGENTS.md
```markdown
---
name: CEO
title: Chief Executive Officer
reportsTo: null
skills:
- paperclip
---
You are the CEO of Lean Dev Shop. You oversee company strategy, coordinate work across the team, and ensure projects ship on time.
Your responsibilities:
- Review and prioritize work across projects
- Coordinate with the CTO on technical decisions
- Ensure the company goals are being met
```
## agents/cto/AGENTS.md
```markdown
---
name: CTO
title: Chief Technology Officer
reportsTo: ceo
skills:
- code-review
- paperclip
---
You are the CTO of Lean Dev Shop. You lead the engineering team and make technical decisions.
Your responsibilities:
- Set technical direction and architecture
- Review code and ensure quality standards
- Mentor engineers and unblock technical challenges
```
## agents/engineer/AGENTS.md
```markdown
---
name: Engineer
title: Software Engineer
reportsTo: cto
skills:
- code-review
- paperclip
---
You are a software engineer at Lean Dev Shop. You write code, fix bugs, and ship features.
Your responsibilities:
- Implement features and fix bugs
- Write tests and documentation
- Participate in code reviews
```
## teams/engineering/TEAM.md
```markdown
---
name: Engineering
description: Product and platform engineering team
slug: engineering
schema: agentcompanies/v1
manager: ../../agents/cto/AGENTS.md
includes:
- ../../agents/engineer/AGENTS.md
- ../../skills/code-review/SKILL.md
tags:
- engineering
---
The engineering team builds and maintains all software products.
```
## projects/q2-launch/PROJECT.md
```markdown
---
name: Q2 Launch
description: Ship the Q2 product launch
slug: q2-launch
owner: cto
---
Deliver all features planned for the Q2 launch, including the new dashboard and API improvements.
```
## projects/q2-launch/tasks/monday-review/TASK.md
```markdown
---
name: Monday Review
assignee: ceo
project: q2-launch
schedule:
timezone: America/Chicago
startsAt: 2026-03-16T09:00:00-05:00
recurrence:
frequency: weekly
interval: 1
weekdays:
- monday
time:
hour: 9
minute: 0
---
Review the status of Q2 Launch project. Check progress on all open tasks, identify blockers, and update priorities for the week.
Coordinate a full Paperclip release across engineering verification, npm,
GitHub, smoke testing, and announcement follow-up. Use when leadership asks
to ship a release, not merely to discuss versioning.
---
# Release Coordination Skill
Run the full Paperclip maintainer release workflow, not just an npm publish.
This skill coordinates:
- stable changelog drafting via `release-changelog`
- canary verification and publish status from `master`
- Docker smoke testing via `scripts/docker-onboard-smoke.sh`
- manual stable promotion from a chosen source ref
- GitHub Release creation
- website / announcement follow-up tasks
## Trigger
Use this skill when leadership asks for:
- "do a release"
- "ship the release"
- "promote this canary to stable"
- "cut the stable release"
## Preconditions
Before proceeding, verify all of the following:
1.`.agents/skills/release-changelog/SKILL.md` exists and is usable.
2. The repo working tree is clean, including untracked files.
3. There is at least one canary or candidate commit since the last stable tag.
4. The candidate SHA has passed the verification gate or is about to.
5. If manifests changed, the CI-owned `pnpm-lock.yaml` refresh is already merged on `master`.
6. npm publish rights are available through GitHub trusted publishing, or through local npm auth for emergency/manual use.
7. If running through Paperclip, you have issue context for status updates and follow-up task creation.
If any precondition fails, stop and report the blocker.
## Inputs
Collect these inputs up front:
- whether the target is a canary check or a stable promotion
- the candidate `source_ref` for stable
- whether the stable run is dry-run or live
- release issue / company context for website and announcement follow-up
## Step 0 — Release Model
Paperclip now uses a commit-driven release model:
1. every push to `master` publishes a canary automatically
2. canaries use `YYYY.MDD.P-canary.N`
3. stable releases use `YYYY.MDD.P`
4. the middle slot is `MDD`, where `M` is the UTC month and `DD` is the zero-padded UTC day
5. the stable patch slot increments when more than one stable ships on the same UTC date
6. stable releases are manually promoted from a chosen tested commit or canary source commit
7. only stable releases get `releases/vYYYY.MDD.P.md`, git tag `vYYYY.MDD.P`, and a GitHub Release
Critical consequences:
- do not use release branches as the default path
- do not derive major/minor/patch bumps
- do not create canary changelog files
- do not create canary GitHub Releases
## Step 1 — Choose the Candidate
For canary validation:
- inspect the latest successful canary run on `master`
- record the canary version and source SHA
For stable promotion:
1. choose the tested source ref
2. confirm it is the exact SHA you want to promote
3. resolve the target stable version with `./scripts/release.sh stable --date YYYY-MM-DD --print-version`
Useful commands:
```bash
git tag --list 'v*' --sort=-version:refname | head -1
git log --oneline --no-merges
npm view paperclipai@canary version
```
## Step 2 — Draft the Stable Changelog
Stable changelog files live at:
-`releases/vYYYY.MDD.P.md`
Invoke `release-changelog` and generate or update the stable notes only.
Rules:
- review the draft with a human before publish
- preserve manual edits if the file already exists
- keep the filename stable-only
- do not create a canary changelog file
## Step 3 — Verify the Candidate SHA
Run the standard gate:
```bash
pnpm -r typecheck
pnpm test:run
pnpm build
```
If the GitHub release workflow will run the publish, it can rerun this gate. Still report local status if you checked it.
For PRs that touch release logic, the repo also runs a canary release dry-run in CI. That is a release-specific guard, not a substitute for the standard gate.
## Step 4 — Validate the Canary
The normal canary path is automatic from `master` via:
-`.github/workflows/release.yml`
Confirm:
1. verification passed
2. npm canary publish succeeded
3. git tag `canary/vYYYY.MDD.P-canary.N` exists
Useful checks:
```bash
npm view paperclipai@canary version
git tag --list 'canary/v*' --sort=-version:refname | head -5
description:Commit SHA, branch, or tag to publish as stable
required:true
type:string
default:master
stable_date:
description:Enter a UTC date in YYYY-MM-DD format, for example 2026-03-18. Do not enter a version string. The workflow will resolve that date to a stable version such as 2026.318.0, then 2026.318.1 for the next same-day stable.
required:false
type:string
dry_run:
description:Preview the stable release without publishing
We really appreciate both small fixes and thoughtful larger changes.
## Two Paths to Get Your Pull Request Accepted
### Path 1: Small, Focused Changes (Fastest way to get merged)
- Pick **one** clear thing to fix/improve
- Touch the **smallest possible number of files**
- Make sure the change is very targeted and easy to review
- All automated checks pass (including Greptile comments)
- No new lint/test failures
These almost always get merged quickly when they're clean.
### Path 2: Bigger or Impactful Changes
- **First** talk about it in Discord → #dev channel
→ Describe what you're trying to solve
→ Share rough ideas / approach
- Once there's rough agreement, build it
- In your PR include:
- Before / After screenshots (or short video if UI/behavior change)
- Clear description of what & why
- Proof it works (manual testing notes)
- All tests passing
- All Greptile + other PR comments addressed
PRs that follow this path are **much** more likely to be accepted, even when they're large.
## General Rules (both paths)
- Write clear commit messages
- Keep PR title + description meaningful
- One PR = one logical change (unless it's a small related group)
- Run tests locally first
- Be kind in discussions 😄
## Writing a Good PR message
Please include a "thinking path" at the top of your PR message that explains from the top of the project down to what you fixed. E.g.:
### Thinking Path Example 1:
> - Paperclip orchestrates ai-agents for zero-human companies
> - There are many types of adapters for each LLM model provider
> - But LLM's have a context limit and not all agents can automatically compact their context
> - So we need to have an adapter-specific configuration for which adapters can and cannot automatically compact their context
> - This pull request adds per-adapter configuration of compaction, either auto or paperclip managed
> - That way we can get optimal performance from any adapter/provider in Paperclip
### Thinking Path Example 2:
> - Paperclip orchestrates ai-agents for zero-human companies
> - But humans want to watch the agents and oversee their work
> - Human users also operate in teams and so they need their own logins, profiles, views etc.
> - So we have a multi-user system for humans
> - But humans want to be able to update their own profile picture and avatar
> - But the avatar upload form wasn't saving the avatar to the file storage system
> - So this PR fixes the avatar upload form to use the file storage service
> - The benefit is we don't have a one-off file storage for just one aspect of the system, which would cause confusion and extra configuration
Then have the rest of your normal PR message after the Thinking Path.
This should include details about what you did, why you did it, why it matters & the benefits, how we can verify it works, and any risks.
Please include screenshots if possible if you have a visible change. (use something like the [agent-browser skill](https://github.com/vercel-labs/agent-browser/blob/main/skills/agent-browser/SKILL.md) or similar to take screenshots). Ideally, you include before and after screenshots.
Questions? Just ask in #dev — we're happy to help.
[](https://www.star-history.com/?repos=paperclipai%2Fpaperclip&type=date&legend=top-left)
This document indexes every part of the Paperclip codebase that touches the [Agent Companies Specification](docs/companies/companies-spec.md) (`agentcompanies/v1-draft`).
Use it when you need to:
1.**Update the spec** — know which implementation code must change in lockstep.
2.**Change code that involves the spec** — find all related files quickly.
3.**Keep things aligned** — audit whether implementation matches the spec.
---
## 1. Specification & Design Documents
| File | Role |
|---|---|
| `docs/companies/companies-spec.md` | **Normative spec** — defines the markdown-first package format (COMPANY.md, TEAM.md, AGENTS.md, PROJECT.md, TASK.md, SKILL.md), reserved files, frontmatter schemas, and vendor extension conventions (`.paperclip.yaml`). |
| `doc/plans/2026-03-13-company-import-export-v2.md` | Implementation plan for the markdown-first package model cutover — phases, API changes, UI plan, and rollout strategy. |
| `doc/SPEC-implementation.md` | V1 implementation contract; references the portability system and `.paperclip.yaml` sidecar format. |
| `docs/specs/cliphub-plan.md` | Earlier blueprint bundle plan; partially superseded by the markdown-first spec (noted in the v2 plan). |
| `doc/plans/2026-02-16-module-system.md` | Module system plan; JSON-only company template sections superseded by the markdown-first model. |
| `doc/plans/2026-03-14-adapter-skill-sync-rollout.md` | Adapter skill sync rollout; companion to the v2 import/export plan. |
## 2. Shared Types & Validators
These define the contract between server, CLI, and UI.
| File | What it defines |
|---|---|
| `packages/shared/src/types/company-portability.ts` | TypeScript interfaces: `CompanyPortabilityManifest`, `CompanyPortabilityFileEntry`, `CompanyPortabilityEnvInput`, export/import/preview request and result types, manifest entry types for agents, skills, projects, issues, recurring routines, companies. |
| `packages/shared/src/validators/company-portability.ts` | Zod schemas for all portability request/response shapes — used by both server routes and CLI. |
| `server/src/services/routines.ts` | Paperclip routine runtime service. Portability now exports routines as recurring `TASK.md` entries and imports recurring tasks back through this service. |
| `server/src/services/company-export-readme.ts` | Generates `README.md` and Mermaid org-chart for exported company packages. |
Route registration lives in `server/src/app.ts` via `companyRoutes(db, storage)`.
## 5. Server — Tests
| File | Coverage |
|---|---|
| `server/src/__tests__/company-portability.test.ts` | Unit tests for the portability service (export, import, preview, manifest shape, `agentcompanies/v1` version). |
| `server/src/__tests__/company-portability-routes.test.ts` | Integration tests for the portability HTTP endpoints. |
## 6. CLI
| File | Commands |
|---|---|
| `cli/src/commands/client/company.ts` | `company export` — exports a company package to disk (flags: `--out`, `--include`, `--projects`, `--issues`, `--projectIssues`).<br>`company import <fromPathOrUrl>` — imports a company package from a file or folder (flags: positional source path/URL or GitHub shorthand, `--include`, `--target`, `--companyId`, `--newCompanyName`, `--agents`, `--collision`, `--ref`, `--dryRun`).<br>Reads/writes portable file entries and handles `.paperclip.yaml` filtering. |
## 7. UI — Pages
| File | Role |
|---|---|
| `ui/src/pages/CompanyExport.tsx` | Export UI: preview, manifest display, file tree visualization, ZIP archive creation and download. Filters `.paperclip.yaml` based on selection. Shows manifest and README in editor. |
| `ui/src/components/PackageFileTree.tsx` | Reusable file tree component for both import and export. Builds tree from `CompanyPortabilityFileEntry` items, parses frontmatter, shows action indicators (create/update/skip), and maps frontmatter field labels. |
| `ui/src/lib/zip.ts` | ZIP archive creation (`createZipArchive`) and reading (`readZipArchive`) — implements ZIP format from scratch for company packages. CRC32, DOS date/time encoding. |
| `ui/src/lib/zip.test.ts` | Tests for ZIP utilities; exercises round-trip with portability file entries and `.paperclip.yaml` content. |
## 10. UI — API Client
| File | Functions |
|---|---|
| `ui/src/api/companies.ts` | `companiesApi.exportBundle`, `companiesApi.exportPreview`, `companiesApi.exportPackage`, `companiesApi.importPreview`, `companiesApi.importBundle` — typed fetch wrappers for the portability endpoints. |
## 11. Skills & Agent Instructions
| File | Relevance |
|---|---|
| `skills/paperclip/references/company-skills.md` | Reference doc for company skill library workflow — install, inspect, update, assign. Skill packages are a subset of the agent companies spec. |
| `server/src/services/company-skills.ts` | Company skill management service — handles SKILL.md-based imports and company-level skill library. |
| `server/src/services/agent-instructions.ts` | Agent instructions service — resolves AGENTS.md paths for agent instruction loading. |
@@ -19,6 +19,14 @@ That's it. On first start the server:
Data persists across restarts in `~/.paperclip/instances/default/db/`. To reset local dev data, delete that directory.
If you need to apply pending migrations manually, run:
```sh
pnpm db:migrate
```
When `DATABASE_URL` is unset, this command targets the current embedded PostgreSQL instance for your active Paperclip config/instance.
This mode is ideal for local development and one-command installs.
Docker note: the Docker quickstart image also uses embedded PostgreSQL by default. Persist `/paperclip` to keep DB state across container restarts (see `doc/DOCKER.md`).
- Do not commit `pnpm-lock.yaml` in pull requests.
- Pull request CI validates dependency resolution when manifests change.
- Pushes to `master` regenerate `pnpm-lock.yaml` with `pnpm install --lockfile-only --no-frozen-lockfile`, commit it back if needed, and then run verification with `--frozen-lockfile`.
## Start Dev
From repo root:
@@ -29,6 +37,10 @@ This starts:
- API server: `http://localhost:3100`
- UI: served by the API server in dev middleware mode (same origin as API)
`pnpm dev` runs the server in watch mode and restarts on changes from workspace packages (including adapter packages). Use `pnpm dev:once` to run without file watching.
`pnpm dev:once` now tracks backend-relevant file changes and pending migrations. When the current boot is stale, the board UI shows a `Restart required` banner. You can also enable guarded auto-restart in `Instance Settings > Experimental`, which waits for queued/running local agent runs to finish before restarting the dev server.
Tailscale/private-auth dev mode:
```sh
@@ -79,6 +91,10 @@ docker compose -f docker-compose.quickstart.yml up --build
See `doc/DOCKER.md` for API key wiring (`OPENAI_API_KEY` / `ANTHROPIC_API_KEY`) and persistence details.
## Docker For Untrusted PR Review
For a separate review-oriented container that keeps `codex`/`claude` login state in Docker volumes and checks out PRs into an isolated scratch workspace, see `doc/UNTRUSTED-PR-REVIEW.md`.
## Database in Dev (Auto-Handled)
For local development, leave `DATABASE_URL` unset.
@@ -114,6 +130,123 @@ When a local agent run has no resolved project/session workspace, Paperclip fall
This path honors `PAPERCLIP_HOME` and `PAPERCLIP_INSTANCE_ID` in non-default setups.
For `codex_local`, Paperclip also manages a per-company Codex home under the instance root and seeds it from the shared Codex login/config home (`$CODEX_HOME` or `~/.codex`):
When developing from multiple git worktrees, do not point two Paperclip servers at the same embedded PostgreSQL data directory.
Instead, create a repo-local Paperclip config plus an isolated instance for the worktree:
```sh
paperclipai worktree init
# or create the git worktree and initialize it in one step:
pnpm paperclipai worktree:make paperclip-pr-432
```
This command:
- writes repo-local files at `.paperclip/config.json` and `.paperclip/.env`
- creates an isolated instance under `~/.paperclip-worktrees/instances/<worktree-id>/`
- when run inside a linked git worktree, mirrors the effective git hooks into that worktree's private git dir
- picks a free app port and embedded PostgreSQL port
- by default seeds the isolated DB in `minimal` mode from the current effective Paperclip instance/config (repo-local worktree config when present, otherwise the default instance) via a logical SQL snapshot
Seed modes:
-`minimal` keeps core app state like companies, projects, issues, comments, approvals, and auth state, preserves schema for all tables, but omits row data from heavy operational history such as heartbeat runs, wake requests, activity logs, runtime services, and agent session state
-`full` makes a full logical clone of the source instance
-`--no-seed` creates an empty isolated instance
After `worktree init`, both the server and the CLI auto-load the repo-local `.paperclip/.env` when run inside that worktree, so normal commands like `pnpm dev`, `paperclipai doctor`, and `paperclipai db:backup` stay scoped to the worktree instance.
That repo-local env also sets:
-`PAPERCLIP_IN_WORKTREE=true`
-`PAPERCLIP_WORKTREE_NAME=<worktree-name>`
-`PAPERCLIP_WORKTREE_COLOR=<hex-color>`
The server/UI use those values for worktree-specific branding such as the top banner and dynamically colored favicon.
Print shell exports explicitly when needed:
```sh
paperclipai worktree env
# or:
eval"$(paperclipai worktree env)"
```
### Worktree CLI Reference
**`pnpm paperclipai worktree init [options]`** — Create repo-local config/env and an isolated instance for the current worktree.
| Option | Description |
|---|---|
| `--name <name>` | Display name used to derive the instance id |
| `--instance <id>` | Explicit isolated instance id |
| `--home <path>` | Home root for worktree instances (default: `~/.paperclip-worktrees`) |
| `--from-config <path>` | Source config.json to seed from |
| `--from-data-dir <path>` | Source PAPERCLIP_HOME used when deriving the source config |
| `--from-instance <id>` | Source instance id (default: `default`) |
| `--server-port <port>` | Preferred server port |
| `--db-port <port>` | Preferred embedded Postgres port |
**`pnpm paperclipai worktree:make <name> [options]`** — Create `~/NAME` as a git worktree, then initialize an isolated Paperclip instance inside it. This combines `git worktree add` with `worktree init` in a single step.
| Option | Description |
|---|---|
| `--start-point <ref>` | Remote ref to base the new branch on (e.g. `origin/main`) |
| `--instance <id>` | Explicit isolated instance id |
| `--home <path>` | Home root for worktree instances (default: `~/.paperclip-worktrees`) |
| `--from-config <path>` | Source config.json to seed from |
| `--from-data-dir <path>` | Source PAPERCLIP_HOME used when deriving the source config |
| `--from-instance <id>` | Source instance id (default: `default`) |
| `--server-port <port>` | Preferred server port |
| `--db-port <port>` | Preferred embedded Postgres port |
**`pnpm paperclipai worktree env [options]`** — Print shell exports for the current worktree-local Paperclip instance.
| Option | Description |
|---|---|
| `-c, --config <path>` | Path to config file |
| `--json` | Print JSON instead of shell exports |
Examples:
```sh
pnpm paperclipai worktree env
pnpm paperclipai worktree env --json
eval"$(pnpm paperclipai worktree env)"
```
For project execution worktrees, Paperclip can also run a project-defined provision command after it creates or reuses an isolated git worktree. Configure this on the project's execution workspace policy (`workspaceStrategy.provisionCommand`). The command runs inside the derived worktree and receives `PAPERCLIP_WORKSPACE_*`, `PAPERCLIP_PROJECT_ID`, `PAPERCLIP_AGENT_ID`, and `PAPERCLIP_ISSUE_*` environment variables so each repo can bootstrap itself however it wants.
## Quick Health Checks
In another terminal:
@@ -141,6 +274,36 @@ pnpm dev
If you set `DATABASE_URL`, the server will use that instead of embedded PostgreSQL.
## Automatic DB Backups
Paperclip can run automatic DB backups on a timer. Defaults:
Agent env vars now support secret references. By default, secret values are stored with local encryption and only secret refs are persisted in agent config.
@@ -216,5 +379,61 @@ Agent-oriented invite onboarding now exposes machine-readable API docs:
-`GET /api/invites/:token` returns invite summary plus onboarding and skills index links.
-`GET /api/invites/:token/onboarding.txt` returns a plain-text onboarding doc intended for both human operators and agents (llm.txt-style handoff), including optional inviter message and suggested network host candidates.
-`GET /api/skills/index` lists available skill documents.
-`GET /api/skills/paperclip` returns the Paperclip heartbeat skill markdown.
## OpenClaw Join Smoke Test
Run the end-to-end OpenClaw join smoke harness:
```sh
pnpm smoke:openclaw-join
```
What it validates:
- invite creation for agent-only join
- agent join request using `adapterType=openclaw`
- board approval + one-time API key claim semantics
- callback delivery on wakeup to a dockerized OpenClaw-style webhook receiver
Required permissions:
- This script performs board-governed actions (create invite, approve join, wakeup another agent).
- In authenticated mode, run with board auth via `PAPERCLIP_AUTH_HEADER` or `PAPERCLIP_COOKIE`.
Optional auth flags (for authenticated mode):
-`PAPERCLIP_AUTH_HEADER` (for example `Bearer ...`)
-`PAPERCLIP_COOKIE` (session cookie header value)
## OpenClaw Docker UI One-Command Script
To boot OpenClaw in Docker and print a host-browser dashboard URL in one command:
```sh
pnpm smoke:openclaw-docker-ui
```
This script lives at `scripts/smoke/openclaw-docker-ui.sh` and automates clone/build/config/start for Compose-based local OpenClaw UI testing.
Pairing behavior for this smoke script:
- default `OPENCLAW_DISABLE_DEVICE_AUTH=1` (no Control UI pairing prompt for local smoke; no extra pairing env vars required)
- set `OPENCLAW_DISABLE_DEVICE_AUTH=0` to require standard device pairing
Model behavior for this smoke script:
- defaults to OpenAI models (`openai/gpt-5.2` + OpenAI fallback) so it does not require Anthropic auth by default
State behavior for this smoke script:
- defaults to isolated config dir `~/.openclaw-paperclip-smoke`
- resets smoke agent state each run by default (`OPENCLAW_RESET_STATE=1`) to avoid stale provider/auth drift
Networking behavior for this smoke script:
- auto-detects and prints a Paperclip host URL reachable from inside OpenClaw Docker
- default container-side host alias is `host.docker.internal` (override with `PAPERCLIP_HOST_FROM_CONTAINER` / `PAPERCLIP_HOST_PORT`)
- if Paperclip rejects container hostnames in authenticated/private mode, allow `host.docker.internal` via `pnpm paperclipai allowed-hostname host.docker.internal` and restart Paperclip
PAPERCLIP_PORT=3200PAPERCLIP_DATA_DIR=./data/pc docker compose -f docker-compose.quickstart.yml up --build
```
If you change host port or use a non-local domain, set `PAPERCLIP_PUBLIC_URL` to the external URL you will use in browser/auth flows.
## Authenticated Compose (Single Public URL)
For authenticated deployments, set one canonical public URL and let Paperclip derive auth/callback defaults:
```yaml
services:
paperclip:
environment:
PAPERCLIP_DEPLOYMENT_MODE:authenticated
PAPERCLIP_DEPLOYMENT_EXPOSURE:private
PAPERCLIP_PUBLIC_URL:https://desk.koker.net
```
`PAPERCLIP_PUBLIC_URL` is used as the primary source for:
- auth public base URL
- Better Auth base URL defaults
- bootstrap invite URL defaults
- hostname allowlist defaults (hostname extracted from URL)
Granular overrides remain available if needed (`PAPERCLIP_AUTH_PUBLIC_BASE_URL`, `BETTER_AUTH_URL`, `BETTER_AUTH_TRUSTED_ORIGINS`, `PAPERCLIP_ALLOWED_HOSTNAMES`).
Set `PAPERCLIP_ALLOWED_HOSTNAMES` explicitly only when you need additional hostnames beyond the public URL host (for example Tailscale/LAN aliases or multiple private hostnames).
## Claude + Codex Local Adapters in Docker
The image pre-installs:
@@ -66,3 +92,45 @@ Notes:
- Without API keys, the app still runs normally.
- Adapter environment checks in Paperclip will surface missing auth/CLI prerequisites.
## Untrusted PR Review Container
If you want a separate Docker environment for reviewing untrusted pull requests with `codex` or `claude`, use the dedicated review workflow in `doc/UNTRUSTED-PR-REVIEW.md`.
That setup keeps CLI auth state in Docker volumes instead of your host home directory and uses a separate scratch workspace for PR checkouts and preview runs.
## Onboard Smoke Test (Ubuntu + npm only)
Use this when you want to mimic a fresh machine that only has Ubuntu + npm and verify:
-`npx paperclipai onboard --yes` completes
- the server binds to `0.0.0.0:3100` so host access works
- onboard/run banners and startup logs are visible in your terminal
- Persistent data is mounted at `./data/docker-onboard-smoke` by default.
- Container runtime user id defaults to your local `id -u` so the mounted data dir stays writable while avoiding root runtime.
- Smoke script defaults to `authenticated/private` mode so `HOST=0.0.0.0` can be exposed to the host.
- Smoke script defaults host port to `3131` to avoid conflicts with local Paperclip on `3100`.
- Smoke script also defaults `PAPERCLIP_PUBLIC_URL` to `http://localhost:<HOST_PORT>` so bootstrap invite URLs and auth callbacks use the reachable host port instead of the container's internal `3100`.
- In authenticated mode, the smoke script defaults `SMOKE_AUTO_BOOTSTRAP=true` and drives the real bootstrap path automatically: it signs up a real user, runs `paperclipai auth bootstrap-ceo` inside the container to mint a real bootstrap invite, accepts that invite over HTTP, and verifies board session access.
- Run the script in the foreground to watch the onboarding flow; stop with `Ctrl+C` after validation.
- Set `SMOKE_DETACH=true` to leave the container running for automation and optionally write shell-ready metadata to `SMOKE_METADATA_FILE`.
- The image definition is in `Dockerfile.onboard-smoke`.
@@ -94,3 +94,53 @@ Canonical mode design and command expectations live in `doc/DEPLOYMENT-MODES.md`
## Further Detail
See [SPEC.md](./SPEC.md) for the full technical specification and [TASKS.md](./TASKS.md) for the task management data model.
---
Paperclip’s core identity is a **control plane for autonomous AI companies**, centered on **companies, org charts, goals, issues/comments, heartbeats, budgets, approvals, and board governance**. The public docs are also explicit about the current boundaries: **tasks/comments are the built-in communication model**, Paperclip is **not a chatbot**, and it is **not a code review tool**. The roadmap already points toward **easier onboarding, cloud agents, easier agent configuration, plugins, better docs, and ClipMart/ClipHub-style reusable companies/templates**.
## What Paperclip should do vs. not do
**Do**
- Stay **board-level and company-level**. Users should manage goals, orgs, budgets, approvals, and outputs.
- Make the first five minutes feel magical: install, answer a few questions, see a CEO do something real.
- Keep work anchored to **issues/comments/projects/goals**, even if the surface feels conversational.
- Treat **agency / internal team / startup** as the same underlying abstraction with different templates and labels.
- Make outputs first-class: files, docs, reports, previews, links, screenshots.
- Provide **hooks into engineering workflows**: worktrees, preview servers, PR links, external review tools.
- Use **plugins** for edge cases like rich chat, knowledge bases, doc editors, custom tracing.
**Do not**
- Do not make the core product a general chat app. The current product definition is explicitly task/comment-centric and “not a chatbot,” and that boundary is valuable.
- Do not build a complete Jira/GitHub replacement. The repo/docs already position Paperclip as organization orchestration, not focused on pull-request review.
- Do not build enterprise-grade RBAC first. The current V1 spec still treats multi-board governance and fine-grained human permissions as out of scope, so the first multi-user version should be coarse and company-scoped.
- Do not lead with raw bash logs and transcripts. Default view should be human-readable intent/progress, with raw detail beneath.
- Do not force users to understand provider/API-key plumbing unless absolutely necessary. There are active onboarding/auth issues already; friction here is clearly real.
## Specific design goals
1.**Time-to-first-success under 5 minutes**
A fresh user should go from install to “my CEO completed a first task” in one sitting.
2.**Board-level abstraction always wins**
The default UI should answer: what is the company doing, who is doing it, why does it matter, what did it cost, and what needs my approval.
3.**Conversation stays attached to work objects**
“Chat with CEO” should still resolve to strategy threads, decisions, tasks, or approvals.
4.**Progressive disclosure**
Top layer: human-readable summary. Middle layer: checklist/steps/artifacts. Bottom layer: raw logs/tool calls/transcript.
5.**Output-first**
Work is not done until the user can see the result: file, document, preview link, screenshot, plan, or PR.
6.**Local-first, cloud-ready**
The mental model should not change between local solo use and shared/private or public/cloud deployment.
7.**Safe autonomy**
Auto mode is allowed; hidden token burn is not.
8.**Thin core, rich edges**
Put optional chat, knowledge, and special surfaces into plugins/extensions rather than bloating the control plane.
This document covers how to build and publish the `paperclipai` CLI package to npm.
Low-level reference for how Paperclip packages are prepared and published to npm.
## Prerequisites
For the maintainer workflow, use [doc/RELEASING.md](RELEASING.md). This document focuses on packaging internals.
- Node.js 20+
- pnpm 9.15+
- An npm account with publish access to the `paperclipai` package
- Logged in to npm: `npm login`
## Current Release Entry Points
## One-Command Publish
Use these scripts:
The fastest way to publish — bumps version, builds, publishes, restores, commits, and tags in one shot:
- [`scripts/release.sh`](../scripts/release.sh) for canary and stable publish flows
- [`scripts/create-github-release.sh`](../scripts/create-github-release.sh) after pushing a stable tag
- [`scripts/rollback-latest.sh`](../scripts/rollback-latest.sh) to repoint `latest`
- [`scripts/build-npm.sh`](../scripts/build-npm.sh) for the CLI packaging build
Paperclip no longer uses release branches or Changesets for publishing.
## Why the CLI needs special packaging
The CLI package, `paperclipai`, imports code from workspace packages such as:
-`@paperclipai/server`
-`@paperclipai/db`
-`@paperclipai/shared`
- adapter packages under `packages/adapters/`
Those workspace references are valid in development but not in a publishable npm package. The release flow rewrites versions temporarily, then builds a publishable CLI bundle.
The script runs all 6 steps below in order. It requires a clean working tree and an active `npm login` session (unless `--dry-run`). After it finishes, push:
```bash
git push && git push origin v<version>
```
## Manual Step-by-Step
If you prefer to run each step individually:
### Quick Reference
```bash
# Bump version
./scripts/version-bump.sh patch # 0.1.0 → 0.1.1
# Build
./scripts/build-npm.sh
# Preview what will be published
cd cli && npm pack --dry-run
# Publish
cd cli && npm publish --access public
# Restore dev package.json
mv cli/package.dev.json cli/package.json
```
## Step-by-Step
This script:
### 1. Bump the version
1. runs the forbidden token check unless `--skip-checks` is supplied
2. runs `pnpm -r typecheck`
3. bundles the CLI entrypoint with esbuild into `cli/dist/index.js`
4. verifies the bundled entrypoint with `node --check`
5. rewrites `cli/package.json` into a publishable npm manifest and stores the dev copy as `cli/package.dev.json`
6. copies the repo `README.md` into `cli/README.md` for npm metadata
After the release script exits, the dev manifest and temporary files are restored automatically.
This updates the versionin two places:
## Package discovery and versioning
-`cli/package.json` — the source of truth
-`cli/src/index.ts` — the Commander `.version()` call
Public packages are discovered from:
-`packages/`
-`server/`
-`cli/`
`ui/` is ignored because it is private.
The version rewrite step now uses [`scripts/release-package-map.mjs`](../scripts/release-package-map.mjs), which:
- finds all public packages
- sorts them topologically by internal dependencies
- rewrites each package version to the target release version
- rewrites internal `workspace:*` dependency references to the exact target version
- updates the CLI's displayed version string
Those rewrites are temporary. The working tree is restored after publish or dry-run.
## Version formats
Paperclip uses calendar versions:
- stable: `YYYY.MDD.P`
- canary: `YYYY.MDD.P-canary.N`
Examples:
```bash
./scripts/version-bump.sh patch # 0.1.0 → 0.1.1
./scripts/version-bump.sh minor # 0.1.0 → 0.2.0
./scripts/version-bump.sh major # 0.1.0 → 1.0.0
./scripts/version-bump.sh 1.2.3 # set explicit version
```
- stable: `2026.318.0`
- canary: `2026.318.1-canary.2`
### 2. Build
## Publish model
### Canary
Canaries publish under the npm dist-tag `canary`.
Example:
-`paperclipai@2026.318.1-canary.2`
This keeps the default install path unchanged while allowing explicit installs with:
```bash
./scripts/build-npm.sh
npx paperclipai@canary onboard
```
The build script runs five steps:
### Stable
1.**Forbidden token check** — scans tracked files for tokens listed in `.git/hooks/forbidden-tokens.txt`. If the file is missing (e.g. on a contributor's machine), the check passes silently. The script never prints which tokens it's searching for.
2.**TypeScript type-check** — runs `pnpm -r typecheck` across all workspace packages.
3.**esbuild bundle** — bundles the CLI entry point (`cli/src/index.ts`) and all workspace package code (`@paperclipai/*`) into a single file at `cli/dist/index.js`. External npm dependencies (express, postgres, etc.) are kept as regular imports.
4.**Generate publishable package.json** — replaces `cli/package.json` with a version that has real npm dependency ranges instead of `workspace:*` references (see [package.dev.json](#packagedevjson) below).
5.**Summary** — prints the bundle size and next steps.
Stable publishes use the npm dist-tag `latest`.
To skip the forbidden token check (e.g. in CI without the token list):
Example:
-`paperclipai@2026.318.0`
Stable publishes do not create a release commit. Instead:
- package versions are rewritten temporarily
- packages are published from the chosen source commit
- git tag `vYYYY.MDD.P` points at that original commit
## Trusted publishing
The intended CI model is npm trusted publishing through GitHub OIDC.
- trusted publisher rules are configured per workflow file
See [doc/RELEASE-AUTOMATION-SETUP.md](RELEASE-AUTOMATION-SETUP.md) for the GitHub/npm setup steps.
## Rollback model
Rollback does not unpublish anything.
It repoints the `latest` dist-tag to a prior stable version:
```bash
./scripts/build-npm.sh --skip-checks
./scripts/rollback-latest.sh 2026.318.0
```
### 3. Preview (optional)
This is the fastest way to restore the default install path if a stable release is bad.
See what npm will publish:
## Related Files
```bash
cd cli &&npmpack --dry-run
```
### 4. Publish
```bash
cd cli && npm publish --access public
```
### 5. Restore dev package.json
After publishing, restore the workspace-aware `package.json`:
```bash
mv cli/package.dev.json cli/package.json
```
### 6. Commit and tag
```bash
git add cli/package.json cli/src/index.ts
git commit -m "chore: bump version to X.Y.Z"
git tag vX.Y.Z
```
## package.dev.json
During development, `cli/package.json` contains `workspace:*` references like:
```json
{
"dependencies":{
"@paperclipai/server":"workspace:*",
"@paperclipai/db":"workspace:*"
}
}
```
These tell pnpm to resolve those packages from the local monorepo. This is great for development but **npm doesn't understand `workspace:*`** — publishing with these references would cause install failures for users.
The build script solves this with a two-file swap:
1.**Before building:**`cli/package.json` has `workspace:*` refs (the dev version).
2.**During build (`build-npm.sh` step 4):**
- The dev `package.json` is copied to `package.dev.json` as a backup.
-`generate-npm-package-json.mjs` reads every workspace package's `package.json`, collects all their external npm dependencies, and writes a new `cli/package.json` with those real dependency ranges — no `workspace:*` refs.
3.**After publishing:** you restore the dev version with `mv package.dev.json package.json`.
The generated publishable `package.json` looks like:
```json
{
"name":"paperclipai",
"version":"0.1.0",
"bin":{"paperclipai":"./dist/index.js"},
"dependencies":{
"express":"^5.1.0",
"postgres":"^3.4.5",
"commander":"^13.1.0"
}
}
```
`package.dev.json` is listed in `.gitignore` — it only exists temporarily on disk during the build/publish cycle.
## How the bundle works
The CLI is a monorepo package that imports code from `@paperclipai/server`, `@paperclipai/db`, `@paperclipai/shared`, and several adapter packages. These workspace packages don't exist on npm.
**esbuild** bundles all workspace TypeScript code into a single `dist/index.js` file (~250kb). External npm packages (express, postgres, zod, etc.) are left as normal `import` statements — they get installed by npm when a user runs `npx paperclipai onboard`.
The esbuild configuration lives at `cli/esbuild.config.mjs`. It automatically reads every workspace package's `package.json` to determine which dependencies are external (real npm packages) vs. internal (workspace code to bundle).
## Forbidden token enforcement
The build process includes the same forbidden-token check used by the git pre-commit hook. This catches any accidentally committed tokens before they reach npm.
claude --print --output-format stream-json --verbose --dangerously-skip-permissions --model claude-opus-4-6 "Use the release-changelog skill to draft or update releases/v${VERSION}.md for Paperclip. Read doc/RELEASING.md and .agents/skills/release-changelog/SKILL.md, then generate the stable changelog for v${VERSION} from commits since the last stable tag. Do not create a canary changelog."
```
The repo intentionally does not run this through GitHub Actions because:
- canaries are too frequent
- stable notes are the only public narrative surface that needs LLM help
- maintainer LLM tokens should not live in Actions
-`.paperclip.yaml` sidecar for Paperclip-specific fidelity
-canonical base package is vendor-neutral and aligned with `docs/companies/companies-spec.md`
-common conventions:
-`agents/<slug>/AGENTS.md`
-`teams/<slug>/TEAM.md`
-`projects/<slug>/PROJECT.md`
-`projects/<slug>/tasks/<slug>/TASK.md`
-`tasks/<slug>/TASK.md`
-`skills/<slug>/SKILL.md`
Export/import behavior in V1:
- export includes company metadata and/or agents based on selection
-export strips environment-specific paths (`cwd`, local instruction file paths)
-export never includes secret values; secret requirements are reported
- export emits a clean vendor-neutral markdown package plus `.paperclip.yaml`
-projects and starter tasks are opt-in export content rather than default package content
-recurring `TASK.md` entries use `recurring: true` in the base package and Paperclip routine fidelity in `.paperclip.yaml`
- Paperclip imports recurring task packages as routines instead of downgrading them to one-time issues
- export strips environment-specific paths (`cwd`, local instruction file paths, inline prompt duplication) while preserving portable project repo/workspace metadata such as `repoUrl`, refs, and workspace-policy references keyed in `.paperclip.yaml`
- export never includes secret values; env inputs are reported as portable declarations instead
- import supports target modes:
- create a new company
- import into an existing company
- import recreates exported project workspaces and remaps portable workspace keys back to target-local workspace ids
- import forces imported agent timer heartbeats off so packages never start scheduled runs implicitly
| `openclaw_gateway` | OpenClaw gateway API | Managed OpenClaw agent via gateway |
| `hermes_local` | Local Hermes process | Hermes agent heartbeat worker |
The `process` and `http` adapters ship as defaults. Additional adapters can be added via the plugin system (see Plugin / Extension Architecture).
The `process` and `http` adapters ship as generic defaults. Additional built-in adapters cover common local coding runtimes (see list above), and new adapter types can be registered via the plugin system (see Plugin / Extension Architecture).
@@ -403,7 +410,7 @@ No separate "agent API" vs. "board API." Same endpoints, different authorization
### Work Artifacts
Paperclip does **not** manage work artifacts (code repos, file systems, deployments, documents). That's entirely the agent's domain. Paperclip tracks tasks and costs. Where and how work gets done is outside scope.
Paperclip manages task-linked work artifacts: issue documents (rich-text plans, specs, notes attached to issues) and file attachments. Agents read and write these through the API as part of normal task execution. Full delivery infrastructure (code repos, deployments, production runtime) remains the agent's domain — Paperclip orchestrates the work, not the build pipeline.
### Open Questions
@@ -429,7 +436,7 @@ The core Paperclip system must be extensible. Features like knowledge bases, ext
- **Agent Adapter plugins** — new Adapter types can be registered via the plugin system
- Plugin-registrable UI components (future)
This isn't a V1 deliverable (we're not building a plugin framework upfront), but the architecture should not paint us into a corner. Keep boundaries clean so extensions are possible.
The plugin framework has shipped. Plugins can register new adapter types, hook into lifecycle events, and contribute UI components (e.g. global toolbar buttons). A plugin SDK and CLI commands (`paperclipai plugin`) are available for authoring and installing plugins.
---
@@ -473,15 +480,14 @@ Each is a distinct page/route:
- [ ]**Default agent** — basic Claude Code/Codex loop with Paperclip skill
- [ ]**Default CEO** — strategic planning, delegation, board communication
- [ ]**Paperclip skill (SKILL.md)** — teaches agents to interact with the API
- [ ]**REST API** — full API for agent interaction (Hono)
- [ ]**REST API** — full API for agent interaction (Express)
- **Does not manage work artifacts** — no repo management, no deployment, no file systems
- **Does not manage delivery infrastructure** — no repo management, no deployment, no file systems (but does manage task-linked documents and attachments)
- **Does not auto-reassign work** — stale tasks are surfaced, not silently redistributed
- **Does not track external revenue/expenses** — that's a future plugin. Token/LLM cost budgeting is core.
docker compose -f docker-compose.untrusted-review.yml run --rm --service-ports review
```
That opens an interactive shell in the review container with:
- Node + Corepack/pnpm
-`codex`
-`claude`
-`gh`
-`git`, `rg`, `fd`, `jq`
## First-time login inside the container
Run these once. The resulting login state persists in the `review-home` Docker volume.
```sh
gh auth login
codex login
claude login
```
If you prefer API-key auth instead of CLI login, pass keys through Compose env:
```sh
OPENAI_API_KEY=... ANTHROPIC_API_KEY=... docker compose -f docker-compose.untrusted-review.yml run --rm review
```
## Check out a PR safely
Inside the container:
```sh
review-checkout-pr paperclipai/paperclip 432
cd /work/checkouts/paperclipai-paperclip/pr-432
```
What this does:
1. Creates or reuses a repo clone under `/work/repos/...`
2. Fetches `pull/<pr>/head` from GitHub
3. Creates a detached git worktree under `/work/checkouts/...`
The checkout lives entirely inside the container volume.
## Ask Codex or Claude to review it
Inside the PR checkout:
```sh
codex
```
Then give it a prompt like:
```text
Review this PR as hostile input. Focus on security issues, data exfiltration paths, sandbox escapes, dangerous install/runtime scripts, auth changes, and subtle behavioral regressions. Do not modify files. Produce findings ordered by severity with file references.
```
Or with Claude:
```sh
claude
```
## Preview the Paperclip app from the PR
Only do this when you intentionally want to execute the PR's code inside the container.
Inside the PR checkout:
```sh
pnpm install
HOST=0.0.0.0 pnpm dev
```
Open from the host:
-`http://localhost:3100`
The Compose file also exposes Vite's default port:
-`http://localhost:5173`
Notes:
-`pnpm install` can run untrusted lifecycle scripts from the PR. That is why this happens inside the isolated container instead of on your host.
- If you only want static inspection, do not run install/dev commands.
- Paperclip's embedded PostgreSQL and local storage stay inside the container home volume via `PAPERCLIP_HOME=/home/reviewer/.paperclip-review`.
## Reset state
Remove the review container volumes when you want a clean environment:
```sh
docker compose -f docker-compose.untrusted-review.yml down -v
```
That deletes:
- Codex/Claude/GitHub login state stored in `review-home`
- cloned repos, worktrees, installs, and scratch data stored in `review-work`
## Security limits
This is a useful isolation boundary, but it is still Docker, not a full VM.
- A reviewed PR can still access the container's network unless you disable it.
- Any secrets you pass into the container are available to code you execute inside it.
- Do not mount your host repo, host home, `.ssh`, or Docker socket unless you are intentionally weakening the boundary.
- If you need a stronger boundary than this, use a disposable VM instead of Docker.
This document summarizes the memory systems referenced in task `PAP-530` and extracts the design patterns that matter for Paperclip.
## What Paperclip Needs From This Survey
Paperclip is not trying to become a single opinionated memory engine. The more useful target is a control-plane memory surface that:
- stays company-scoped
- lets each company choose a default memory provider
- lets specific agents override that default
- keeps provenance back to Paperclip runs, issues, comments, and documents
- records memory-related cost and latency the same way the rest of the control plane records work
- works with plugin-provided providers, not only built-ins
The question is not "which memory project wins?" The question is "what is the smallest Paperclip contract that can sit above several very different memory systems without flattening away the useful differences?"
## Quick Grouping
### Hosted memory APIs
-`mem0`
-`supermemory`
-`Memori`
These optimize for a simple application integration story: send conversation/content plus an identity, then query for relevant memory or user context later.
### Agent-centric memory frameworks / memory OSes
-`MemOS`
-`memU`
-`EverMemOS`
-`OpenViking`
These treat memory as an agent runtime subsystem, not only as a search index. They usually add task memory, profiles, filesystem-style organization, async ingestion, or skill/resource management.
### Local-first memory stores / indexes
-`nuggets`
-`memsearch`
These emphasize local persistence, inspectability, and low operational overhead. They are useful because Paperclip is local-first today and needs at least one zero-config path.
## Per-Project Notes
| Project | Shape | Notable API / model | Strong fit for Paperclip | Main mismatch |
|---|---|---|---|---|
| [nuggets](https://github.com/NeoVertex1/nuggets) | local memory engine + messaging gateway | topic-scoped HRR memory with `remember`, `recall`, `forget`, fact promotion into `MEMORY.md` | good example of lightweight local memory and automatic promotion | very specific architecture; not a general multi-tenant service |
| [mem0](https://github.com/mem0ai/mem0) | hosted + OSS SDK | `add`, `search`, `getAll`, `get`, `update`, `delete`, `deleteAll`; entity partitioning via `user_id`, `agent_id`, `run_id`, `app_id` | closest to a clean provider API with identities and metadata filters | provider owns extraction heavily; Paperclip should not assume every backend behaves like mem0 |
| [MemOS](https://github.com/MemTensor/MemOS) | memory OS / framework | unified add-retrieve-edit-delete, memory cubes, multimodal memory, tool memory, async scheduler, feedback/correction | strong source for optional capabilities beyond plain search | much broader than the minimal contract Paperclip should standardize first |
| [supermemory](https://github.com/supermemoryai/supermemory) | hosted memory + context API | `add`, `profile`, `search.memories`, `search.documents`, document upload, settings; automatic profile building and forgetting | strong example of "context bundle" rather than raw search results | heavily productized around its own ontology and hosted flow |
| [memU](https://github.com/NevaMind-AI/memU) | proactive agent memory framework | file-system metaphor, proactive loop, intent prediction, always-on companion model | good source for when memory should trigger agent behavior, not just retrieval | proactive assistant framing is broader than Paperclip's task-centric control plane |
| [Memori](https://github.com/MemoriLabs/Memori) | hosted memory fabric + SDK wrappers | registers against LLM SDKs, attribution via `entity_id` + `process_id`, sessions, cloud + BYODB | strong example of automatic capture around model clients | wrapper-centric design does not map 1:1 to Paperclip's run / issue / comment lifecycle |
| [EverMemOS](https://github.com/EverMind-AI/EverMemOS) | conversational long-term memory system | MemCell extraction, structured narratives, user profiles, hybrid retrieval / reranking | useful model for provenance-rich structured memories and evolving profiles | focused on conversational memory rather than generalized control-plane events |
| [memsearch](https://github.com/zilliztech/memsearch) | markdown-first local memory index | markdown as source of truth, `index`, `search`, `watch`, transcript parsing, plugin hooks | excellent baseline for a local built-in provider and inspectable provenance | intentionally simple; no hosted service semantics or rich correction workflow |
| [OpenViking](https://github.com/volcengine/OpenViking) | context database | filesystem-style organization of memories/resources/skills, tiered loading, visualized retrieval trajectories | strong source for browse/inspect UX and context provenance | treats "context database" as a larger product surface than Paperclip should own |
## Common Primitives Across The Landscape
Even though the systems disagree on architecture, they converge on a few primitives:
-`ingest`: add memory from text, messages, documents, or transcripts
-`query`: search or retrieve memory given a task, question, or scope
-`scope`: partition memory by user, agent, project, process, or session
-`provenance`: carry enough metadata to explain where a memory came from
-`maintenance`: update, forget, dedupe, compact, or correct memories over time
-`context assembly`: turn raw memories into a prompt-ready bundle for the agent
If Paperclip does not expose these, it will not adapt well to the systems above.
## Where The Systems Differ
These differences are exactly why Paperclip needs a layered contract instead of a single hard-coded engine.
### 1. Who owns extraction?
-`mem0`, `supermemory`, and `Memori` expect the provider to infer memories from conversations.
-`memsearch` expects the host to decide what markdown to write, then indexes it.
-`MemOS`, `memU`, `EverMemOS`, and `OpenViking` sit somewhere in between and often expose richer memory construction pipelines.
Paperclip should support both:
- provider-managed extraction
- Paperclip-managed extraction with provider-managed storage / retrieval
### 2. What is the source of truth?
-`memsearch` and `nuggets` make the source inspectable on disk.
- hosted APIs often make the provider store canonical.
- filesystem-style systems like `OpenViking` and `memU` treat hierarchy itself as part of the memory model.
Paperclip should not require a single storage shape. It should require normalized references back to Paperclip entities.
### 3. Is memory just search, or also profile and planning state?
-`mem0` and `memsearch` center search and CRUD.
-`supermemory` adds user profiles as a first-class output.
-`MemOS`, `memU`, `EverMemOS`, and `OpenViking` expand into tool traces, task memory, resources, and skills.
Paperclip should make plain search the minimum contract and richer outputs optional capabilities.
### 4. Is memory synchronous or asynchronous?
- local tools often work synchronously in-process.
- larger systems add schedulers, background indexing, compaction, or sync jobs.
Paperclip needs both direct request/response operations and background maintenance hooks.
## Paperclip-Specific Takeaways
### Paperclip should own these concerns
- binding a provider to a company and optionally overriding it per agent
- mapping Paperclip entities into provider scopes
- provenance back to issue comments, documents, runs, and activity
- cost / token / latency reporting for memory work
- browse and inspect surfaces in the Paperclip UI
- governance on destructive operations
### Providers should own these concerns
- extraction heuristics
- embedding / indexing strategy
- ranking and reranking
- profile synthesis
- contradiction resolution and forgetting logic
- storage engine details
### The control-plane contract should stay small
Paperclip does not need to standardize every feature from every provider. It needs:
- a required portable core
- optional capability flags for richer providers
- a way to record provider-native ids and metadata without pretending all providers are equivalent internally
## Recommended Direction
Paperclip should adopt a two-layer memory model:
1.`Memory binding + control plane layer`
Paperclip decides which provider key is in effect for a company, agent, or project, and it logs every memory operation with provenance and usage.
2.`Provider adapter layer`
A built-in or plugin-supplied adapter turns Paperclip memory requests into provider-specific calls.
The portable core should cover:
- ingest / write
- search / recall
- browse / inspect
- get by provider record handle
- forget / correction
- usage reporting
Optional capabilities can cover:
- profile synthesis
- async ingestion
- multimodal content
- tool / resource / skill memory
- provider-native graph browsing
That is enough to support:
- a local markdown-first baseline similar to `memsearch`
- hosted services similar to `mem0`, `supermemory`, or `Memori`
- richer agent-memory systems like `MemOS` or `OpenViking`
without forcing Paperclip itself to become a monolithic memory engine.
> Supersession note: the company-template/package-format direction in this document is no longer current. For the current markdown-first company import/export plan, see `doc/plans/2026-03-13-company-import-export-v2.md` and `docs/companies/companies-spec.md`.
## Overview
Paperclip's module system lets you extend the control plane with new capabilities — revenue tracking, observability, notifications, dashboards — without forking core. Modules are self-contained packages that register routes, UI pages, database tables, and lifecycle hooks.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.