tyroneross/code-audit

Full thirty-principle audit with a validated evidence report. Use only when the user explicitly asks for a complete principles audit; for normal reviews use the code-principles skill.

Review, audit, debug or improve code using shared engineering principles and incident memory. Use for code reviews, recurring bugs, inefficiencies, root-cause repairs and pre-build checks.

Checklist for writing and reviewing code that is correct, fast, efficient and scalable. Use when implementing or reviewing non-trivial code, optimizing a slow path, adding remote calls/retries/queues/concurrency/caching, handling errors, or changing a contract other code depends on. Scans the code for signals, loads only the matching checklist sections, and logs whether each finding held up.

Investigate and repair code failures by tracing causal mechanisms, consulting incident memory when available and verifying supported behavior.

Use for nontrivial, uncertain, recurrent or disputed root causes, a repair that did not hold, or an explicit deep investigation. Memory labels inform investigation but do not determine depth.

Consult stored debugging incidents before fresh investigation when the bundled memory MCP is available; validate past causes against current code and record evidence-supported outcomes.

Multi-batch, resumable review campaign across a large repository. Use only when the user explicitly asks for a repository-wide campaign; for a targeted speed fix use the code-principles skill.

Use when the user asks to "add logging", "add tracing", "improve observability", "OpenTelemetry", "structured logging", or reports silent failures or no runtime visibility. Generates stack-appropriate logging with optional OTel.

Structured thirty-principle plan packet before a large change. Use only when the user explicitly asks for a principles plan; otherwise apply the code-principles checklist while planning.