Bug Fix Skill

Reproduce and analyze bugs, propose options, and apply a safe selected fix.

<when_to_use>

Situation	Example
Type errors	Property 'X' does not exist on type 'Y'
Runtime errors	Cannot read property ... of undefined
Logic bugs	duplicate rendering, wrong calculation, stale state
API failures	4xx/5xx, response shape mismatch
Intermittent issues	failures under specific conditions

</when_to_use>

<argument_validation>

If ARGUMENT is missing, ask immediately:

Which bug should be fixed?
- Error message / stack trace
- Expected vs actual behavior
- Reproduction steps
- Related files (if known)

</argument_validation>

<mandatory_reasoning>

Mandatory Sequential Thinking

• Always run sequential-thinking before implementation.
• Simple bug: at least 3 thoughts
• Medium complexity: 5 thoughts
• High complexity: 7+ thoughts

Recommended sequence:

1. Complexity classification
2. Reproduction and symptom framing
3. Root-cause hypotheses
4. Option comparison
5. Final recommendation

</mandatory_reasoning>

Step	Task	Tool
1	Validate input and summarize reproducible signals	-
2	Build analysis plan with sequential-thinking	sequential-thinking
3	Explore relevant code and isolate root cause	Read/Grep/Glob
4	Present 2-3 fix options	-
5	Wait for user selection	-
6	Implement selected option	Edit
7	Run type/test/build verification	Bash
8	Report outcome and changed files	-

<option_presentation>

Use this format:

## Bug Analysis Result
Root cause: ...
Impact scope: ...

### Option 1: ... (Recommended)
- Pros:
- Cons:
- Risk:
- Files:

### Option 2: ...
- Pros:
- Cons:
- Risk:
- Files:

### Option 3: ... (Temporary)
- Pros:
- Cons:
- Risk:
- Files:

Recommendation: Option N (reason ...)
Which option should I apply? (1/2/3)

</option_presentation>

<implementation_rules>

• Do not modify code before user option selection.
• Avoid speculative edits; use evidence-based fixes only.
• Keep scope limited to the requested bug and direct impact.
• Always include validation output in the final report.

</implementation_rules>

Execution checklist:

• ARGUMENT validated
• sequential-thinking completed
• Root-cause evidence collected
• 2-3 options presented
• User choice confirmed
• typecheck/test/build executed
• outcome + touched files reported

Forbidden:

• speculative fix without evidence
• immediate implementation without options
• implementation without explicit user choice
• completion claim without validation