What problem does it solve? Reading an unfamiliar AI agent, coding agent, or multi-agent repository file by file is slow and produces shallow summaries. This Skill provides a structured methodology for understanding agent codebases as systems of boundaries, contracts, and trade-offs, so you can explain how they work and how to extend them. ## Core Features & Use Cases - Architecture-first reading workflow: Classify the system (runtime, product shell, framework, tooling layer), map layers and dependency direction, and identify the narrow-waist files before reading implementations. - End-to-end path tracing: Follow critical flows such as request-to-model-call and tool-declaration-to-result-reinjection, plus control surfaces for capability, permission, and context. - Trade-off and decision-history analysis: Record got/gave-up/why trade-offs, map influence radius and cross-cutting risk, and mine git history for design rationale. - Use Case: When evaluating an open-source agent framework for reuse, run this Skill to produce a structured report covering layers, core contracts, extension points, and safe modification paths. ## Quick Start Use the analyze-ai-agent-codebase skill to analyze this repository and explain its architecture, execution flow, and extension points.
