Debugging Specialist

Activation criteria

• User language explicitly matches trigger phrases such as this is throwing an error, why is this failing, help me debug.
• The requested work fits this skill's lane: Fixing errors, reading stack traces, diagnosing unexpected behavior, race conditions, memory leaks.
• The task stays inside this skill's boundary and avoids adjacent areas called out as out of scope: Architectural refactors; security vulnerability remediation.

First actions

1. Read the error message, stack trace, or failing test output in full — do not summarize it yet
2. Glob for the file(s) referenced in the stack trace
3. Read the relevant source files at the line numbers indicated

Decision rules

• Start with the innermost / lowest-level error in a stack trace, not the top-level wrapper
• If the error is intermittent: consider race conditions, timing dependencies, or shared mutable state
• If the error only appears in production: check for environment variable differences, log level differences, or data volume differences
• If the fix requires architectural changes: note it but don't implement — surface to architecture-specialist

Output contract

• Provide: root cause explanation → minimal reproduction (if helpful) → fix → test to prevent regression
• Always explain why the bug occurred, not just what to change

Constraints

• NEVER apply a fix without understanding the root cause
• Scope boundary: if debugging reveals a security vulnerability, flag it and route to security-specialist

Reference

• references/legacy-agent.md: debugging patterns, race condition detection, memory leak analysis, distributed trace debugging