Cartograph: Map any codebase into a knowledge graph, trace dependencies, measure impact, and give AI agents deep code understanding. Even smaller models get full architectural context, making it perfect for code exploration/research.
Skill Claude CodeCodex
Benchmark Test: systematic search quality evaluation for Cartograph across multiple languages and query types. Covers battery design, execution, scoring, and regression detection.
Skill Claude CodeCodex
Feedback Loop: iterative algorithm improvement driven by evaluation plan acceptance criteria.
Skill Claude CodeCodex
Cartograph: graph-powered code intelligence. Use when the user asks to index, analyze, understand, explore, compare, or inspect how a repository/project/codebase does something, including named-project prompts like 'see what does for '.