What problem does it solve?
Traditional version control systems like Git struggle with multiple AI agents making simultaneous changes, leading to frequent conflicts, lock-waiting, and slow context switching. Agentic Jujutsu provides a solution designed for AI-native, concurrent development.
Core Features & Use Cases
- Self-Learning Version Control: Tracks operations, learns patterns, and provides intelligent suggestions using ReasoningBank.
- Quantum-Resistant Security: Ensures fast integrity verification and future-proof protection with quantum-resistant cryptography.
- Multi-Agent Coordination: Enables multiple AI agents to work concurrently without conflicts, offering 23x faster concurrent commits than Git.
- Automated Conflict Resolution: Achieves an 87% success rate in automatically resolving conflicts, minimizing manual intervention.
- Use Case: Coordinate a swarm of AI agents to simultaneously implement, test, and review a new feature, with Agentic Jujutsu handling all versioning, learning from successful trajectories, and resolving conflicts automatically.
Quick Start
Install Agentic Jujutsu:
npx agentic-jujutsu
Create a new commit:
const { JjWrapper } = require('agentic-jujutsu');
const jj = new JjWrapper();
await jj.newCommit('Add feature');
Start a self-learning trajectory:
const id = jj.startTrajectory('Implement authentication');
