Files
msd-core/gsd-core/templates/DEBUG.md
Tom Boucher 5e52350736 feat(#1958): add multi-signal fix-acceptance guardrail to gsd-debugger (#2396)
* test(#1958): add failing-first guardrail contract tests

Epic #1957 Phase 1A. Adds source-text-is-the-product tests asserting the
5-signal fix-acceptance guardrail contract (target test, mutation check,
no-op/deletion detector, adjacent tests, revert-and-reconfirm), graceful
degradation, FIX REJECTED BY GUARDRAIL return path, per-signal debug-file
recording, and subprocess bounding.

Failing-first: reference file and agent sections do not yet exist.

* feat(#1958): add multi-signal fix-acceptance guardrail to gsd-debugger

Epic #1957 Phase 1A. Prevents accepting a fix that merely greens the test
(Goodhart defense / APR overfitting). Adds a 5-signal gate run before fix
acceptance: target test, mutation check (Stryker), no-op/behavior-deleting
detector, adjacent/held-out tests, revert-and-reconfirm. Degrades gracefully
when Stryker or a test suite is absent (each skip logged, never a silent pass),
records per-signal results under Resolution.verification, and returns a
FIX REJECTED BY GUARDRAIL outcome the session-manager surfaces for
revise / accept-as-debt / abandon.

Full rules extracted to gsd-core/references/debugger-fix-acceptance.md (slim
routing kept in the agent to respect the agent-size cap). Debug template +
INVENTORY + manifest + agent-size baseline + AGENTS.md updated.

* test(#1958): correct newline-tolerant assertion + regen install-parity goldens

The revert-and-reconfirm assertion collapsed whitespace before matching so
markdown line-wrapping does not break it. Regenerated the golden-install-parity
and install-tree fixtures (npm run gen:golden) to absorb the intentional
gsd-debugger.md / gsd-debug-session-manager.md / DEBUG.md / new reference-file
changes to the installed artifact tree.

* fix(#1958): tighten guardrail per orthogonal review

Addresses the isolated reviewer's findings:
- signal 5 now states its recorded-repro dependency and routes the no-repro
  case to the degradation row; revert mechanism specified (git stash / git
  revert -n); minimality flag tied to diff structure, not revert-ability.
- bounded-subprocesses section now bounds the git subprocess (5-30s) too,
  requires argv-array argument passing, and scopes Stryker to the driving
  regression test (a mutant killed only by a non-driving test is a finding).
- new test-provenance (security) clause: the driving test must be
  agent-authored; bug-report repro scripts are DATA, never executed verbatim.
- tightened 3 contract assertions to bind to specific clauses
  (guardrail_verdict field, deletion-reject-unless-RCA, 60s+git bounding).
- Goodhart framing softened to 'partially-independent'; DEBUG.md template
  verification field notes the nested map shape.

* chore(#1958): backfill changeset pr number (PR #2396)

* fix(#1958): add issue ref to allow-test-rule annotation (ADR-456)

CI lint-allow-test-rule-refs requires every allow-test-rule exemption to
carry a 'see #NNN' issue ref per ADR-456. The new test file's annotation
lacked it; this adds (see #1958).
2026-07-18 00:39:28 -04:00

170 lines
5.3 KiB
Markdown

# Debug Template
Template for `.planning/debug/[slug].md` — active debug session tracking.
---
## File Template
```markdown
---
status: gathering | investigating | fixing | verifying | awaiting_human_verify | resolved
trigger: "[verbatim user input]"
created: [ISO timestamp]
updated: [ISO timestamp]
---
## Current Focus
<!-- OVERWRITE on each update - always reflects NOW -->
hypothesis: [current theory being tested]
test: [how testing it]
expecting: [what result means if true/false]
next_action: [immediate next step — be specific, not "continue investigating"]
reasoning_checkpoint: null <!-- populated before every fix attempt — see structured_returns -->
tdd_checkpoint: null <!-- populated when tdd_mode is active after root cause confirmed -->
## Symptoms
<!-- Written during gathering, then immutable -->
expected: [what should happen]
actual: [what actually happens]
errors: [error messages if any]
reproduction: [how to trigger]
started: [when it broke / always broken]
## Eliminated
<!-- APPEND only - prevents re-investigating after /clear -->
- hypothesis: [theory that was wrong]
evidence: [what disproved it]
timestamp: [when eliminated]
## Evidence
<!-- APPEND only - facts discovered during investigation -->
- timestamp: [when found]
checked: [what was examined]
found: [what was observed]
implication: [what this means]
## Resolution
<!-- OVERWRITE as understanding evolves -->
root_cause: [empty until found]
fix: [empty until applied]
verification: [empty until verified — holds the nested per-signal fix-acceptance guardrail record (map shape) when active; see gsd-core/references/debugger-fix-acceptance.md]
files_changed: []
```
---
<section_rules>
**Frontmatter (status, trigger, timestamps):**
- `status`: OVERWRITE - reflects current phase
- `trigger`: IMMUTABLE - verbatim user input, never changes
- `created`: IMMUTABLE - set once
- `updated`: OVERWRITE - update on every change
**Current Focus:**
- OVERWRITE entirely on each update
- Always reflects what Claude is doing RIGHT NOW
- If Claude reads this after /clear, it knows exactly where to resume
- Fields: hypothesis, test, expecting, next_action, reasoning_checkpoint, tdd_checkpoint
- `next_action`: must be concrete and actionable — bad: "continue investigating"; good: "Add logging at line 47 of auth.js to observe token value before jwt.verify()"
- `reasoning_checkpoint`: OVERWRITE before every fix_and_verify — five-field structured reasoning record (hypothesis, confirming_evidence, falsification_test, fix_rationale, blind_spots)
- `tdd_checkpoint`: OVERWRITE during TDD red/green phases — test file, name, status, failure output
**Symptoms:**
- Written during initial gathering phase
- IMMUTABLE after gathering complete
- Reference point for what we're trying to fix
- Fields: expected, actual, errors, reproduction, started
**Eliminated:**
- APPEND only - never remove entries
- Prevents re-investigating dead ends after context reset
- Each entry: hypothesis, evidence that disproved it, timestamp
- Critical for efficiency across /clear boundaries
**Evidence:**
- APPEND only - never remove entries
- Facts discovered during investigation
- Each entry: timestamp, what checked, what found, implication
- Builds the case for root cause
**Resolution:**
- OVERWRITE as understanding evolves
- May update multiple times as fixes are tried
- Final state shows confirmed root cause and verified fix
- Fields: root_cause, fix, verification, files_changed
</section_rules>
<lifecycle>
**Creation:** Immediately when /gsd:debug is called
- Create file with trigger from user input
- Set status to "gathering"
- Current Focus: next_action = "gather symptoms"
- Symptoms: empty, to be filled
**During symptom gathering:**
- Update Symptoms section as user answers questions
- Update Current Focus with each question
- When complete: status → "investigating"
**During investigation:**
- OVERWRITE Current Focus with each hypothesis
- APPEND to Evidence with each finding
- APPEND to Eliminated when hypothesis disproved
- Update timestamp in frontmatter
**During fixing:**
- status → "fixing"
- Update Resolution.root_cause when confirmed
- Update Resolution.fix when applied
- Update Resolution.files_changed
**During verification:**
- status → "verifying"
- Update Resolution.verification with results
- If verification fails: status → "investigating", try again
**After self-verification passes:**
- status -> "awaiting_human_verify"
- Request explicit user confirmation in a checkpoint
- Do NOT move file to resolved yet
**On resolution:**
- status → "resolved"
- Move file to .planning/debug/resolved/ (only after user confirms fix)
</lifecycle>
<resume_behavior>
When Claude reads this file after /clear:
1. Parse frontmatter → know status
2. Read Current Focus → know exactly what was happening
3. Read Eliminated → know what NOT to retry
4. Read Evidence → know what's been learned
5. Continue from next_action
The file IS the debugging brain. Claude should be able to resume perfectly from any interruption point.
</resume_behavior>
<size_constraint>
Keep debug files focused:
- Evidence entries: 1-2 lines each, just the facts
- Eliminated: brief - hypothesis + why it failed
- No narrative prose - structured data only
If evidence grows very large (10+ entries), consider whether you're going in circles. Check Eliminated to ensure you're not re-treading.
</size_constraint>