ONE-SCAN REPO CONTEXT GENERATION WITH CODEBASE-MD
Codebase-md scans your repo and auto-generates consistent AI coding context files for popular tools, reducing manual drift and improving prompt quality across t...
Codebase-md scans your repo and auto-generates consistent AI coding context files for popular tools, reducing manual drift and improving prompt quality across teams.
The tool inspects languages, frameworks, architecture, dependencies, and git history, then emits formats like CLAUDE.md and .cursorrules from a single source model (DEV Community, Forem mirror). It uses tree-sitter ASTs to infer conventions and parses package manifests to flag version freshness and potential breaking changes.
Smart context routing chunks your project into 12 topics and ranks results via TF-IDF to answer targeted queries like “database models,” ready to feed into any LLM. A plugin-style generator layer and 300+ tests suggest it’s designed for extension and CI use.
Usage is simple: pip install, scan the repo, then generate formats in one go. This helps standardize AI assistant inputs across Claude Code, Cursor, Windsurf, multi-agent setups, and more without hand-maintaining multiple files.
Unified, auto-generated context files cut config drift and speed AI-assisted coding.
Deeper project analysis improves retrieval relevance and reduces noisy prompts.
-
terminal
Run scan+generate on large monorepos to gauge runtime, relevance, and noise.
-
terminal
Validate that outputs exclude secrets and internal-only paths before CI rollout.
Legacy codebase integration strategies...
- 01.
Adopt in dry-run mode and diff against existing CLAUDE.md or .cursorrules before replacing.
- 02.
Gate generation by path filters to avoid vendor folders and massive binaries.
Fresh architecture paradigms...
- 01.
Template new services with codebase-md in CI to regenerate context on each merge.
- 02.
Standardize team prompts around generated files to align coding assistants from day one.