mirror of https://github.com/glittercowboy/get-shit-done synced 2026-04-26 01:35:29 +02:00

Files

Tom Boucher 18bb0149c8 feat: add workflow.discuss_mode assumptions config (#637 )

Add codebase-first assumption-driven alternative to the interview-style
discuss-phase. New `workflow.discuss_mode: "assumptions"` config routes
to a separate workflow that spawns a gsd-assumptions-analyzer agent to
read 5-15 codebase files, surface assumptions with evidence, and ask
only for corrections (~2-4 interactions vs ~15-20).

- New gsd-assumptions-analyzer agent for deep codebase analysis
- New discuss-phase-assumptions.md workflow (15 steps)
- Command-level routing via dual @reference + process gate
- Identical CONTEXT.md output — downstream agents unaffected
- Existing discuss-phase.md workflow untouched (zero diff)
- Mode-aware plan-phase gate and progress display
- User documentation and integration tests
- Update agent count and list in copilot-install tests (17 → 18)

Closes #637

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-03-21 00:21:17 -04:00

2.6 KiB

Raw Blame History

Discuss Mode: Assumptions vs Interview

GSD's discuss-phase has two modes for gathering implementation context before planning.

Modes

`discuss` (default)

The original interview-style flow. Claude identifies gray areas in the phase, presents them for selection, then asks ~4 questions per area. Good for:

Early phases where the codebase is new
Phases where the user has strong opinions they want to express proactively
Users who prefer guided, conversational context gathering

`assumptions`

A codebase-first flow. Claude deeply analyzes the codebase via a subagent (reading 5-15 relevant files), forms assumptions with evidence, and presents them for confirmation or correction. Good for:

Established codebases with clear patterns
Users who find the interview questions obvious
Faster context gathering (~2-4 interactions vs ~15-20)

Configuration

# Enable assumptions mode
gsd-tools config-set workflow.discuss_mode assumptions

# Switch back to interview mode
gsd-tools config-set workflow.discuss_mode discuss

The setting is per-project (stored in .planning/config.json).

How Assumptions Mode Works

Init — Same as discuss mode (load prior context, scout codebase, check todos)
Deep analysis — Explore subagent reads 5-15 codebase files related to the phase
Surface assumptions — Each assumption includes:
- What Claude would do and why (citing file paths)
- What goes wrong if the assumption is incorrect
- Confidence level (Confident / Likely / Unclear)
Confirm or correct — User reviews assumptions, selects any that need changing
Write CONTEXT.md — Identical output format to discuss mode

Flag Compatibility

Flag	`discuss` mode	`assumptions` mode
`--auto`	Auto-selects recommended answers	Skips confirm gate, auto-resolves Unclear items
`--batch`	Groups questions in batches	N/A (corrections already batched)
`--text`	Plain-text questions (remote sessions)	Plain-text questions (remote sessions)
`--analyze`	Shows trade-off tables per question	N/A (assumptions include evidence)

Output

Both modes produce identical CONTEXT.md with the same 6 sections:

<domain> — Phase boundary
<decisions> — Locked implementation decisions
<canonical_refs> — Specs/docs downstream agents must read
<code_context> — Reusable assets, patterns, integration points
<specifics> — User references and preferences
<deferred> — Ideas noted for future phases

Downstream agents (researcher, planner, checker) consume this identically regardless of mode.

2.6 KiB Raw Blame History