humansys/raise/rai-bugfix-analyse
Claude Code and agent skill by humansys/raise
- community
Checking that the SKILL.md file is still in the repository.
Root cause analysis using the method best suited to the bug. Phase 3 of bugfix pipeline.
Install
npx vskill@latest install humansys/raise/rai-bugfix-analyseSKILL.md
Bugfix Analyse
Purpose
Determine the root cause of the bug and decide a fix approach. Select the analysis method based on what information is available — not on an arbitrary complexity tier.
Mastery Levels (ShuHaRi)
- Shu: Follow the selected method strictly, document every hypothesis
- Ha: Combine methods when signals overlap
- Ri: Domain-specific RCA methods (fault trees, SBFL)
Context
When to use: After /rai-bugfix-triage has classified the bug in 4 dimensions.
When to skip: XS bugs where cause is self-evident — but still write analysis.md documenting the obvious cause.
Inputs: Bug ID, work/bugs/{issue_key}/scope.md with TRIAGE block.
Expected state: On bug branch. Scope with triage exists. Bug reproduces.
Steps
Step 1: Select Method Based on Available Signals
Read the scope artifact and determine which signals you have:
| Signal available | Best method | |
|---|---|---|
| Stack trace or error message with location | Stack trace analysis — follow the error path to origin | |
| Bug appeared after a known change | git bisect — binary search commits to find the introducing change | |
| Single suspected cause | 5 Whys — trace one causal chain, each answer evidenced | |
| Multiple possible causes, unclear which | Hypothesis-driven — list hypotheses, test each, eliminate | |
| Cause is evident from reproduction | Document directly — write the cause and move on |
Choose the method that matches your signals. If uncertain, default to hypothesis-driven — it's the most general and what LLMs do naturally.
Step 2: Analyse
Execute the selected method:
Stack trace analysis: Follow the error from the surface (exception/log) back through the call chain to the originating defect. Document the path.
git bisect: git bisect start, git bisect bad HEAD, git bisect good {known-good-commit}. Test at each step until the introducing commit is found.
5 Whys: Trace one factual causal chain — ask "Why?" up to five times, each answer evidenced by code. Stop at the actionable root cause.
Hypothesis-driven: List 2-5 hypotheses for the root cause. For each one, define a test (grep, read code, run command), execute it, and record the result:
| Hypothesis | Test | Result | Conclusion | |
|---|---|---|---|---|
| {what might cause it} | {how to verify} | {what you found} | confirmed / eliminated |
Narrow to one confirmed root cause. If two remain, escalate to human at GATE 2.
Rule for all methods: "Human error" is never a root cause — ask why the error was possible.
<verification> Root cause stated with evidence. Fix approach decided — not implemented yet. </verification>
<if-blocked> Root cause unclear after analysis → document top 2 hypotheses, escalate to human at GATE 2. </if-blocked>
Step 3: Write Analysis Artifact
Publish work/bugs/{issue_key}/analysis.md via CLI with: method used, root cause, evidence, and fix approach:
Use raise_docs_write MCP tool with doc_type="bugfix-analysis", title="{issue_key}: analysis", content="## Method: {method used}\n\n### Root Cause\n{root cause}\n\n### Evidence\n{evidence supporting root cause}\n\n### Fix Approach\n{one line}", output_path="work/bugs/{issue_key}/analysis.md", cwd="{project_or_worktree_path}". If MCP tools are not available, fall back to:
<verification> Analysis artifact committed with method, root cause, and fix approach. </verification>
Output
| Item | Destination | |
|---|---|---|
| Analysis | work/bugs/{issue_key}/analysis.md |
STOP HERE. Return your summary to the orchestrator. Do NOT invoke any further skill.
Quality Checklist
- [ ] Method selected based on available signals (not arbitrary tier)
- [ ] Root cause confirmed with evidence (not guessed)
- [ ] Fix approach decided but not implemented
- [ ] Analysis artifact committed
- [ ] "Human error" never accepted as root cause
- [ ] NEVER fix before analysing — symptoms recur without root cause
References
- Previous:
/rai-bugfix-triage - Next:
/rai-bugfix-plan
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
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-analyse/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-analyse)