debug-agent

Debug Specialist and Root Cause Analyst. Activate Diana when the user needs help investigating a bug, performing systematic code inspection, analyzing root causes, validating a fix, or running any structured debugging workflow. Diana uses Fagan inspection (60-90% defect detection rate), fishbone analysis, binary search debugging, and 13 specialized workflows. Use this agent for anything related to debugging, defect analysis, code quality issues, or production incidents — even if the user just describes a problem without explicitly asking for a debug session.

Identify missing assertions, preconditions, postconditions, and invariants in code. Use to add defensive programming that catches bugs early, makes code self-documenting, and surfaces violations before they cause hard-to-debug failures downstream. Trigger for: "add defensive checks", "make this code fail fast", "harden this module", "I want to catch invalid state early", or during code review when you want to improve robustness.

Reduce a complex failing test case to the smallest possible reproduction that still triggers the bug. Use when a bug requires complex setup to reproduce, when you want to create a clean bug report, or when the reproduction case is too noisy to debug directly. Trigger for: "I need a minimal reproduction", "the test to reproduce this is 500 lines", "can you simplify this failing case", or before filing a bug report with a library or framework.

Comprehensive 6-phase Fagan inspection for systematic defect analysis — achieves 60-90% defect detection rates. Use for complex, critical, or hard-to-reproduce bugs that need thorough investigation. Trigger for: "I need a thorough bug analysis", P0/P1 incidents, bugs that keep coming back, "we've tried everything and can't find it", pre-release code audits, or any situation where a quick fix isn't enough.

Design a strategic logging, metrics, and tracing plan for production debugging and observability. Use when production issues are hard to diagnose because there aren't enough logs, when setting up monitoring for a new service, or when existing logs are too noisy or not useful. Trigger for: "we can't tell what's happening in production", "add logging to this service", "design observability for X", "we need better metrics", or after a production incident where lack of visibility was a problem.

Analyze commit history and codebase for systemic defect patterns, hotspots, and recurring issues. Use when multiple related bugs are appearing, when a component seems fragile, or when you want to understand the health of a codebase before a release. Trigger for: "we keep getting bugs in this area", "this module feels flaky", "analyze our recent issues", pre-release quality audits, or when you want to find problems before users do.