Semantic
intelligence
for real codebases.
AST-aware analysis, semantic Git diffing, and AI-friendly export — all running locally with zero telemetry.
Capabilities
Everything you need.
Nothing you don't.
AST Analysis
Deep semantic understanding of your codebase using precise Abstract Syntax Trees. Parses Python, TypeScript, JavaScript, and more.
Semantic Diffing
Understand what actually changed logically, ignoring whitespace and formatting noise. See intent, not lines.
Export Engine
Pack your entire repository into LLM-optimized formats instantly. Markdown, JSON, or custom schemas — ready for any model.
Explain Engine
Generate intelligent, context-aware explanations for complex legacy modules. Powered by local semantic graphs.
Diagnostics
Catch structural anomalies and architectural decay before they hit production. Circular deps, layer violations, and more.
Local-first Workflows
Zero telemetry. All intelligence and processing runs securely on your local machine. No cloud, no data leaks.
Architecture
How it works.
From raw code to semantic intelligence.
Diagnostics
Catch problems early.
Before they reach production.
brain project doctor validates your entire environment — from Git availability to API key presence — and reports a clear status.
- brain project doctor
Checks project init, Git, config, and provider setup
- brain project summary
Total files, functions, classes, architecture hints
- brain diff review
Semantic diff explanations with optional LLM — HTML reports
- brain diff explain <file>
Explain a specific file or function: src/api.py:create_user
Motivation
Why project-brain exists.
Fixing modern development workflows.
Git diffs are noisy
Traditional diffs show line changes, not semantic meaning. A small refactor can generate large diffs while hiding the actual behavioral impact.
Codebases become difficult to understand
Large repositories contain deeply nested modules, duplicated logic, and hidden dependencies. Understanding them manually is slow and error-prone.
AI tools require structured context
Most AI systems perform poorly when fed raw repositories. project-brain creates AI-friendly context and function-level intelligence to maximize output quality.
Local-first tooling matters
Many developers do not want automatic code uploads or vendor lock-in. project-brain works fully offline with no cloud dependency — your code stays yours.
CLI
Designed for your terminal.
Fast, intuitive, and pipeable.
A clean, composable CLI — both brain and project-brain aliases work.
30-second quick start
Requires Python 3.10+ and Git.View full documentation on GitHub →