ultrathink-detective

AST-based multi-perspective audits examine architecture, implementation, and testing layers simultaneously. This Skill uses map, symbol, callers, and callees commands to reveal health issues, dependency patterns, and technical debt across complex codebases without grep or glob, producing architecture maps and exact impact locations.

Yes. Multi-perspective AST analysis reconstructs full call chains, semantic relationships, and architectural context to identify root causes. It's designed for post-incident analysis, revealing how changes propagate through systems and exposing interconnected issues spanning multiple components.

AST audits combine architect, developer, tester, and debugger perspectives in one holistic assessment, producing dependency tracing and call-chain evidence. Standard reviews typically cover single perspectives. This approach surfaces architectural debt, testing gaps, and implementation risks that isolated reviews miss.

Yes. This requires claudemem v0.3.0 in AST mode with access to map, symbol, callers, callees, context, and search commands. The environment must support semantic code analysis and dependency mapping without relying on grep or find operations.

Use comprehensive audits for major refactors, security audits, architecture decisions, complex bugs spanning systems, and technical debt assessment. Targeted reviews work for isolated changes; holistic audits reveal interconnected health issues, testing gaps, and systemic risks that point reviews don't surface.

Multi-perspective analysis examines architecture (design violations), implementation (code complexity), and testing (coverage gaps) simultaneously using semantic symbol search and call-chain tracing. This reveals debt sources—poor abstractions, untested paths, tight coupling—and quantifies their health impact across the system.