thkt/fix

/fix - Quick Bug Fix

Rapidly fix small bugs with root cause analysis and TDD verification.

Input

$ARGUMENTS holds a bug description, a finding ID from a /audit snapshot in ${CLAUDE_SKILL_DIR}/../../workspace/history/ (e.g., RC-001, SEC-003), or a finding returned by a standalone audit workflow run. Scope is limited to small, well-understood issues of 1-3 files. The $ARGUMENTS pattern routes the mode.

| Pattern | Mode | Action | | --------------------------------------------- | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | /^[A-Z]+-[0-9]+$/ | Finding ID Resolution | Read the latest snapshot from ${CLAUDE_SKILL_DIR}/../../workspace/history/ and find the ID match in findings[]. Carry severity / fix_type / root cause, skip Outcome Anchor and Build Check, enter Triage. If absent, present error + suggest Standard Flow | | Finding with file / line / severity / summary | Direct Finding Input | Return value of the audit workflow: a single JSON finding, or text carrying file:line + severity + summary. Use file:line as the RCA starting point, skip Outcome Anchor and Build Check, enter Triage | | empty | Fix Prompt | Ask via AskUserQuestion for Fix type from Bug fix / Error message / Test failure and Description as free text via Other, then execute | | otherwise | Standard Flow | Treat as a bug description and run from Outcome Anchor |

When Direct Finding Input carries multiple findings, handle them one at a time in descending severity order. When the impact spans 4+ files, check the multi-file trigger in § Escalation first.

Delegation Map

| Type | Target | Purpose | | --------- | ---------------------------------------------------- | ------------------------------------------------- | | Skill | use-context-root-cause-analysis | 5 Whys for non-obvious bugs | | Agent | generator-test | Regression test from symptom + repro steps | | Agent | resolver-build | TypeScript or build error triage | | Reference | ${CLAUDE_SKILL_DIR}/references/defense-in-depth.md | Multi-layer validation for Recurring / Systematic |

Outcome Anchor

Read .claude/OUTCOME.md before Build Check. If absent, generate the stub via /outcome. Confirm the bug or fix lives inside the outcome state. If outside, § Escalation.

Build Check

Detect the build command from package.json or project config and run it.

| Result | Action | | ------------ | ----------------------------------------------------- | | Build errors | Task with subagent_type: resolver-build, then END | | No errors | Continue to Triage |

Triage

Obvious skips both RCA and regression test generation, so it is limited to findings with low misfix risk.

| Input | Condition | Path | | -------------------- | -------------------------------------------------------------- | ----------- | | Bug desc | Single location identified + 1-3 line fix + no similar pattern | Obvious | | Bug desc | Intermittent, multiple repro conditions, or unknown root cause | Non-obvious | | Finding ID | fix_type: auto and severity low / med | Obvious | | Finding ID | severity critical / high, or fix_type not auto | Non-obvious | | Direct Finding Input | severity low / med and a 1-3 line fix | Obvious | | Direct Finding Input | severity critical / high, or the fix is non-obvious | Non-obvious |

Obvious

1. Apply minimal fix
2. Run tests covering affected code if any exist

Non-obvious

1. Run 5 Whys via Skill("use-context-root-cause-analysis"). If via Finding ID or Direct Finding Input, pass the finding's file:line and summary as the 5 Whys starting point, adding the snapshot root cause when available. Output Symptom / Root cause / Pattern.
2. Task(subagent_type: generator-test) for the regression test. Pass symptom + repro steps only
3. Verify regression test is Red
4. Apply fix
5. Verify regression test is Green and no other tests regressed
6. If Pattern is Recurring or Systematic, apply ${CLAUDE_SKILL_DIR}/references/defense-in-depth.md

Escalation

Branch on objective triggers, not confidence self-assessment. Do not attempt fix #4 without escalating.

| Trigger | Action | | ------------------------------ | ----------------------------------------------------------------------- | | RCA cannot identify root cause | Escalate to /research | | Tests still fail after fix | Re-analyze root cause. After 3 failures, escalate to /research | | Multi-file impact (4+ files) | Refine via /issue and delegate to the build workflow | | New feature scope | Refine via /issue and delegate to the build workflow | | Pattern = Systematic | Escalate to /research | | Fix outside OUTCOME.md scope | Confirm with user; redefine Non-goals or delegate to the build workflow |

Error Handling

| Error | Action | | ---------------------- | -------------------------------------- | | resolver-build fails | Present error, ask user for guidance | | generator-test timeout | Skip regression test, proceed with fix |

Verification

| Check | Required | | ---------------------------------------- | --------------------------------------- | | Root cause identified (Non-obvious path) | Yes | | All tests pass | Yes | | Pattern field recorded from RCA | Yes (Non-obvious path) | | defense-in-depth applied if needed | Yes (Recurring / Systematic only) | | Re-audit suggested if via finding | Yes (Finding ID / Direct Finding Input) |