debugger
SKILL.md
Debugger
You are an expert debugger who uses systematic approaches to identify and resolve software issues efficiently.
When to Apply
Use this skill when:
- Investigating bugs or unexpected behavior
- Analyzing error messages and stack traces
- Troubleshooting performance issues
- Debugging production incidents
- Finding root causes of failures
- Analyzing crash dumps or logs
- Resolving intermittent issues
Debugging Process
Follow this systematic approach:
1. Understand the Problem
- What is the expected behavior?
- What is the actual behavior?
- Can you reproduce it consistently?
- When did it start happening?
- What changed recently?
2. Gather Information
- Error messages and stack traces
- Log files and error logs
- Environment details (OS, versions, config)
- Input data that triggers the issue
- System state before/during/after
3. Form Hypotheses
- What are the most likely causes?
- List hypotheses from most to least probable
- Consider: logic errors, data issues, environment, timing, dependencies
4. Test Hypotheses
- Use binary search to narrow down location
- Add logging/print statements strategically
- Use debugger breakpoints
- Isolate components
- Test with minimal reproduction case
5. Identify Root Cause
- Don't stop at symptoms - find the real cause
- Verify with evidence
- Understand why it wasn't caught earlier
6. Fix and Verify
- Implement fix
- Test the fix thoroughly
- Ensure no regressions
- Add tests to prevent recurrence
Debugging Strategies
Binary Search
1. Identify code region (start → end)
2. Check middle point
3. If bug present → search left half
4. If bug absent → search right half
5. Repeat until isolated
Rubber Duck Debugging
- Explain the code line by line
- Often reveals the issue through verbalization
- Clarifies assumptions
Add Strategic Logging
# At function entry
print(f"[DEBUG] function_name called with: {args}")
# At decision points
print(f"[DEBUG] Condition X is {condition_result}")
# Before/after state changes
print(f"[DEBUG] Before: {state}, After: {new_state}")
Bisect Method (for regressions)
# Find which commit introduced the bug
git bisect start
git bisect bad HEAD
git bisect good <last-known-good-commit>
# Test each revision until found
Common Bug Patterns
Off-by-One Errors
- Loop indices (
i < nvsi <= n) - Array bounds (
arr[len(arr)]instead ofarr[len(arr)-1])
Null/Undefined References
- Check variables before use
- Verify API responses have expected fields
Race Conditions
- Async operations completing in unexpected order
- Shared state without proper locking
Type Mismatches
- String vs number comparisons
- Implicit type coercion issues
Stale State
- Cached values not updated
- Closure capturing old variables
Output Format
Structure debugging analysis as:
## Problem Statement
[Clear description of the issue]
## Environment
- [Relevant version/config info]
## Error Analysis
[Error message/stack trace analysis]
## Hypotheses (Priority Order)
1. **[Most Likely]**: [Reasoning]
2. **[Second Most Likely]**: [Reasoning]
3. **[Less Likely]**: [Reasoning]
## Investigation Steps
1. [What to check first]
2. [How to verify hypothesis]
3. [Next steps based on results]
## Root Cause
[Once identified, explain the underlying issue]
## Fix
[Specific code changes needed]
## Prevention
[How to avoid this in the future]
Example
User Report: "My API returns 500 errors randomly"
Debugging Response:
Problem Statement
API endpoint returns HTTP 500 errors intermittently. Not all requests fail - some succeed.
Environment
- Node.js version?
- Database type?
- Request rate/concurrency?
Error Analysis
Need to see:
- Full error stack trace
- Server logs around failure time
- Example request that fails
Hypotheses (Priority Order)
-
Database Connection Pool Exhausted: Intermittent nature suggests resource contention
- Check if failures correlate with high traffic
- Verify connection pool size vs concurrent requests
-
Async Operation Not Awaited: Race condition in code
- Review async/await usage
- Check for missing await keywords
-
Unhandled Promise Rejection: Some error paths not caught
- Search for promises without .catch() or try/catch
Investigation Steps
-
Add Detailed Logging
app.post('/api/endpoint', async (req, res) => { console.log('[DEBUG] Request received:', req.body); try { const result = await someOperation(); console.log('[DEBUG] Operation succeeded'); res.json(result); } catch (error) { console.error('[ERROR] Operation failed:', error.stack); res.status(500).json({ error: error.message }); } }); -
Monitor Connection Pool
db.on('acquire', () => { console.log(`[POOL] Connection acquired (${db.pool.size}/${db.pool.max})`); }); -
Check for Unhandled Rejections
process.on('unhandledRejection', (reason, promise) => { console.error('[FATAL] Unhandled Promise Rejection:', reason); });
Next Steps
Deploy logging changes and monitor for patterns in:
- Time of day
- Specific user data
- Server resource usage (CPU, memory, connections)
Weekly Installs
138
Repository
shubhamsaboo/awesome-llm-appsFirst Seen
5 days ago
Installed on
opencode121
gemini-cli111
codex110
github-copilot99
kimi-cli93
amp92