debug

<user_request> $ARGUMENTS </user_request>

<critical_override> THIS COMMAND OVERRIDES THE CLAUDE.md TASK ROUTING TABLE FOR AGENT SELECTION.

WHY: The CLAUDE.md routing table maps "Investigation" to code-analysis:detective, but this command needs dev:debugger for error analysis and dev:developer for applying fixes. code-analysis:detective is READ-ONLY and cannot fix bugs.

AGENT RULES FOR THIS COMMAND:

• Error analysis and root cause investigation → dev:debugger agent (subagent_type: "dev:debugger")
• Applying fixes → dev:developer agent (subagent_type: "dev:developer")
• Stack detection → dev:stack-detector agent (subagent_type: "dev:stack-detector")
• Validation → Bash (run tests directly)

DO NOT use code-analysis:detective (READ-ONLY — cannot apply fixes or write code). DO NOT use dev:researcher (researches topics, does NOT debug code). DO NOT use dev:architect (plans architecture, does NOT debug or fix code). </critical_override>

  Before starting, create todo list with all 6 phases:
  0. Initialize (understand issue)
  1. Error Analysis
  2. Root Cause Investigation
  3. Fix Implementation
  4. Validation
  5. Documentation

  Update continuously as you progress.
</todowrite_requirement>

<orchestrator_role>
  **You are an ORCHESTRATOR, not IMPLEMENTER.**

  **You MUST:**
  - Use Task tool to delegate ALL debugging work
  - Use Tasks to track investigation steps
  - Document findings systematically
  - Trace root cause before fixing

  **You MUST NOT:**
  - Write or edit code directly
  - Apply fixes yourself
  - Skip root cause analysis
</orchestrator_role>

<delegation_rules>
  - ALL error analysis -> debugger agent
  - ALL fixes -> developer agent
  - ALL validation -> run tests via Bash
</delegation_rules>

</critical_constraints>

      Detect technology stack for this project.
      Save results to: ${SESSION_PATH}/context.json
      ```
    </step>
    <step>Mark PHASE 0 as completed</step>
  </steps>
  <quality_gate>Issue understood, stack detected</quality_gate>
</phase>

<phase number="1" name="Error Analysis">
  <objective>Analyze error and identify potential causes</objective>
  <steps>
    <step>Mark PHASE 1 as in_progress</step>
    <step>
      Launch debugger agent:
      ```
      SESSION_PATH: ${SESSION_PATH}
      DETECTED_STACK: {stack}

      Analyze this error:
      {error_description}

      Steps:
      1. Parse stack trace/error message
      2. Identify error type and category
      3. List potential root causes (ranked by likelihood)
      4. Identify relevant code files
      5. Suggest investigation paths

      Save analysis to: ${SESSION_PATH}/error-analysis.md
      ```
    </step>
    <step>Review analysis (read ${SESSION_PATH}/error-analysis.md)</step>
    <step>Mark PHASE 1 as completed</step>
  </steps>
  <quality_gate>Error analyzed, potential causes identified</quality_gate>
</phase>