humansys/raise/rai-bugfix-review
Claude Code and agent skill by humansys/raise
- community
Checking that the SKILL.md file is still in the repository.
Retrospective, pattern extraction, and process improvement. Phase 6 of bugfix pipeline.
Install
npx vskill@latest install humansys/raise/rai-bugfix-reviewSKILL.md
Bugfix Review
Purpose
Verify the fix addresses root cause, extract process improvements and causal patterns, and produce the retrospective artifact. This is where bugs become organizational learning.
Mastery Levels (ShuHaRi)
- Shu: Follow all steps, answer every checkpoint question, add patterns
- Ha: Skip checkpoint for trivial fixes; focus patterns on novel insights
- Ri: Feed systemic findings to graph; cross-bug pattern analysis
Context
When to use: After /rai-bugfix-fix has completed all planned tasks with passing gates.
When to skip: Never — even trivial fixes produce learnings. Skipping review is the #1 step-skipping failure mode.
Inputs: Bug ID, all prior artifacts (scope.md, analysis.md, plan.md), code commits.
Expected state: On bug branch. All tasks committed. All gates pass. Bug no longer reproduces.
Steps
Step 1: Heutagogical Checkpoint
Answer with specific examples:
- What did you learn about this system or codebase?
- What would you change about the fix process?
- Are there improvements for the framework (skill, guardrail, template)?
- What are you more capable of now?
Step 1a: Architecture Review Checklist & Attestation
Run the AR checklist and write the attestation marker that /rai-bugfix-close's gate-ar-bugfix gate verifies. This is the only place the checklist runs — bugfix-close does not re-run it, it only checks that this marker exists (RAISE-14326 / ADR-130: attestation and verification must not be the same actor at the same instant — mirrors the story pipeline's split, RAISE-14277).
- P1 — Structural drift:
rai drift check packages/<changed-module>/
Review output: orphaned symbols? dead public APIs?
- P2 — Beck-R2: Does this fix add necessary complexity only?
No speculative abstractions, no unused parameters, no dead branches.
- P3 — Convention: Naming, module placement, public API surface
consistent with codebase?
After confirming all three, write the branch-and-session-scoped attestation marker via the centralized writer (D1 — never hand-roll this path in bash):
Do not run the gate-ar-bugfix gate here — verification happens later, in /rai-bugfix-close, against this same marker.
<verification> AR checklist (P1/P2/P3) completed. Attestation marker written for /rai-bugfix-close to verify. </verification>
Step 2: Extract Patterns & Process Improvements
Pattern writing guidelines — before composing pattern content:
- Structure: action + context + reason (e.g., "Use singleton for DB connections to avoid pool exhaustion under concurrent requests")
- Length: 100-300 chars ideal. If > 500, consider splitting into 2 patterns
- Be specific: "Validate JWT expiry before DB query" > "Always validate tokens"
- No narrative: state the insight directly, not the story of how you found it
- Bad example: "We found that using a singleton pattern works better because in our testing we saw that multiple connections caused issues"
- Good example: "Use singleton pattern for DB connections to avoid connection pool exhaustion under concurrent requests"
Add patterns worth preserving. Use raise_pattern_add MCP tool with cwd="{project_or_worktree_path}", content="{causal insight}", context="{keywords}", pattern_type="process", from_story="{issue_key}". If MCP tools are not available, fall back to:
Types: process, technical, architecture, codebase.
Reinforce behavioral patterns loaded at session start. Use raise_pattern_reinforce MCP tool with pattern_id={pattern_id}, vote={1|0|-1}, from_story="{issue_key}", cwd="{project_or_worktree_path}". If MCP tools are not available, fall back to:
| Vote | Meaning | |
|---|---|---|
1 | Fix followed the pattern | |
0 | Pattern not relevant (does NOT count toward scoring) | |
-1 | Fix contradicted the pattern |
Process improvement — answer with specifics:
- What change in process or tooling would prevent this class of bug?
- What classification pattern does this bug represent?
Step 3: Write Retrospective
Write work/bugs/{issue_key}/retro.md (Write tool) with this template:
Commit:
<verification> Retro written. Patterns added/reinforced. </verification>
Output
| Item | Destination | |
|---|---|---|
| Retrospective | work/bugs/{issue_key}/retro.md | |
| Patterns | .raise/rai/memory/patterns.jsonl |
STOP HERE. Return your summary to the orchestrator. Do NOT invoke any further skill.
Quality Checklist
- [ ] Heutagogical checkpoint answered with specifics
- [ ] Process improvement extracted with prevention + pattern
- [ ] Patterns added with
--scope projectif applicable - [ ] Retro artifact committed
- [ ] NEVER merge without retro — learnings compound
- [ ] NEVER skip pattern reinforce — scoring system depends on it
References
- Previous:
/rai-bugfix-fix - Next:
/rai-bugfix-close - Pattern scoring: RAISE-170 (temporal decay + Wilson scorer)
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingpackages/raise-cli/src/raise_cli/skills_base/rai-bugfix-review/SKILL.md
Popularity
- Repository stars
- 71
- Repository forks
- 8
- 7-day trend score
- 21
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/humansys/raise/rai-bugfix-review)