Ouroboros
AI-powered code refactoring with discrete diffusion
Get Started CLI Reference View on GitHub
Project Overview
Ouroboros is a production-ready AI code generation system that combines GraphRAG for infinite context, discrete diffusion for high-quality code generation, and comprehensive safety gates with complete provenance logging.
Key Features
- 🛡️ Safety Gates: Tree-Sitter syntax validation prevents invalid code
- 🔄 Self-Healing: Automatic retry loops with error feedback
- 🧠 Discrete Diffusion: High-quality code generation (Phase 4)
- 📊 Complete Provenance: Full audit trail for every run
- 🖥️ Beautiful CLI: Rich terminal interface with Typer
- ⚡ GraphRAG Context: Neo4j knowledge graph for deterministic retrieval
All Phases Complete ✅
| Phase | Status | Description |
|---|---|---|
| Phase 1 | ✅ Complete | The Librarian - Knowledge Graph (Neo4j + Tree-sitter) |
| Phase 2 | ✅ Complete | The Reasoner - Dependency Analysis & Planning |
| Phase 3 | ✅ Complete | The Compressor - Context Encoding (Jamba 256k) |
| Phase 4 | ✅ Complete | The Builder - Discrete Diffusion Code Generation |
| Phase 5 | ✅ Complete | Integration Loop - Safety, CLI, Provenance |
Quick Start
# Install
pip install -r requirements.txt
# Refactor code (mock mode)
python ouroboros_cli.py refactor "Add caching" \
--target src/user_service.py \
--mock \
--dry-run
# Check status
python ouroboros_cli.py status --latest