codebase-intel

What problem does it solve?

Prevent AI agents from making junior-level mistakes by answering code-structure questions and impact/risk checks using deterministic, precomputed repository intelligence rather than ad-hoc guessing.

Core Features & Use Cases

• Codebase Discovery & Q&A: Summaries, architecture overviews, circular/dead code findings, and git history attribution for “what does X do?” and “where is Y implemented?” questions.
• Risk & Blast-Radius Guardrails: File risk scoring (0–100) and dependency-based blast radius analysis to guide safe edits and force explicit approval for high-risk changes.
• Graph Maintenance & Gap Filling: Detects unmapped/low-coverage areas in the structural graph, checks node metadata for gaps, fills edges only above a confidence threshold, and refreshes analysis incrementally.
• Source-based Context: Uses graph.db for structural facts and MASTER.md for architectural context, reducing token waste from rereading raw code.

Quick Start

Ask your agent: “Use codebase-intel to summarize <file_path>, report its risk score, then show the blast radius for the change before I edit anything.”