ultrathink-detective

Ultrathink-detective performs AST-based analysis using claudemem commands—map, symbol, callers, and callees—to reveal codebase structure and call relationships without text-search tools. It synthesizes architecture, security, and implementation perspectives into a unified report with prioritized findings.

Yes. Ultrathink-detective is designed for complex, multi-language repositories during security audits, refactors, and incident investigations. It provides evidence-backed conclusions by combining architecture, testing, debugging, security, and performance analyses in parallel.

Run ultrathink-detective's multi-dimensional analysis—map, symbol, callers, callees, context, and search commands—to uncover structure, call relationships, and architectural gaps. The consensus-driven approach synthesizes findings into actionable remediation steps.

Yes, ultrathink-detective requires claudemem v0.3.0. The Skill uses claudemem's AST commands—map, symbol, callers, callees, context, and search—and prohibits grep, find, and glob usage for strict AST-only analysis.

AST-based analysis requires claudemem v0.3.0 and structured symbol/call-graph queries; it cannot use grep, find, or glob patterns. Trade-off: precision and architectural insight gain at the cost of flexibility in ad-hoc text searches across unstructured content.

Ultrathink-detective combines architecture, implementation, testing, debugging, security, and performance perspectives using claudemem commands. Consensus-driven synthesis across all dimensions produces deeper insights and identifies gaps that single-perspective analysis would miss.