Debugger Agent
Systematic root cause analysis for production incidents, API failures, and complex technical issues.
When to Use
- API endpoints returning 500 errors or unexpected responses
- CI/CD pipeline failures blocking deployments
- Database connection pools exhausted or queries timing out
- Production incidents requiring immediate diagnosis
Key Capabilities
| Area | What It Does |
|---|---|
| Issue Investigation | Structured problem-solving: assess severity, collect logs, identify patterns, trace timeline |
| Database Analysis | Schema inspection (psql \d), query plans (EXPLAIN), connection monitoring, lock detection |
| Log Analysis | Parse server logs, CI/CD output, GitHub Actions failures, container logs, system errors |
| Performance | Response times, resource usage (CPU/memory/disk), bottleneck identification, cache analysis |
| Root Cause | Error tracing, dependency failures, config issues, code bugs, infrastructure problems |
Common Use Cases
Backend Engineer: API 500 Errors
Prompt: /debug [POST /api/orders returning 500, started after v2.3.4 deploy]
Gets root cause (missing req.user, connection leak), fix plan, rollback steps, validation commands.
DevOps: Database Connection Exhaustion
Prompt: /debug [PostgreSQL pool exhausted, 47/20 connections active]
Identifies leaked transactions, long-running queries, table locks. Provides kill commands, query timeouts, code fixes.
Full-Stack Dev: GitHub Actions Failing
Prompt: /debug [CI build failing on test step, error "Module not found"]
Analyzes workflow logs, identifies missing dependency or broken import, suggests package.json fix.
Site Reliability Engineer: Performance Degradation
Prompt: /debug [API latency increased from 200ms to 3s after deploy]
Profiles endpoints, finds N+1 queries or missing indexes, provides EXPLAIN ANALYZE output and optimization plan.
Pro Tips
Collect context first: Before running /debug, gather error messages, timestamps, recent changes (deploys/commits), and environment details.
Check the usual suspects: Recent deployments, config changes, dependency updates, database migrations often cause issues.
Use parallel investigation: Combine with scout agents for broad searches: scout('Find all database transaction usage', 3) while analyzing logs.
Validate fixes in staging: Test proposed solutions in non-production before applying to prod.
Document for next time: Debugger reports become runbooks - save them for similar future incidents.
Related Agents
- Tester - Validate fixes with comprehensive tests
- Code Reviewer - Review fix quality before merge
- Fullstack Developer - Implement suggested fixes
Key Takeaway
Debugger agent systematically investigates technical issues from symptoms to root cause, providing actionable solutions with validation steps and prevention measures - turning hours of troubleshooting into 30-minute structured analysis.