Kungfu aims to solve the problem of AI agents losing context between sessions, which forces users to re-explain projects and agents to reconstruct work. It provides verified project context, durable work continuity, and inspectable understanding, allowing agents to efficiently pick up where they left off. This benefits developers and power users who rely on AI agents for complex project tasks, reducing context pressure and improving the reliability of agent-driven workflows.

How It Works

The system allows agents to inspect declared project sources, building a verified context rather than relying solely on conversational history. This approach makes missing context, conflicts, and decisions visible. Kungfu preserves the agent's understanding, intent, action boundaries, and outcomes across fresh contexts and agent processes, ensuring coherent progress even after conversation history is lost or sessions are restarted.

Quick Start & Requirements

To evaluate the current source tree, clone the repository and build from source: git clone https://github.com/kungfu-systems/kungfu.git cd kungfu ./shifu doctor ./shifu sync && ./shifu build

The intended one-command path for agent interaction is sh kungfu run agent. Public release artifacts are not yet available.

• Product home: https://kungfu.tech
• Developer and agent surface: https://libkungfu.dev

Highlighted Details

• Provides verified project context, enabling agents to select specific context instead of loading entire repositories.
• Ensures durable work continuity, preserving intent, understanding, and outcomes across context loss and restarts.
• Offers inspectable understanding, keeping sources, omissions, and uncertainty visible to users and agents.
• Aims for initial value within minutes, with deeper benefits unfolding over days of sustained use.

Maintenance & Community

Questions and issues can be directed via GitHub issue forms. Security reports should follow SECURITY.md. Specific community channels or contributor details are not detailed in the provided README.

Licensing & Compatibility

The project is licensed under the Apache License 2.0. This license is generally permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

Kungfu v4 is marked as "Coming soon," and public release artifacts are not yet available. Source-built capabilities are primarily for qualification. Features such as cross-platform evidence, strong power-loss durability, and a robust institutional profile remain staged. Users should be aware that design intent, implemented behavior, qualified guarantees, and released artifacts are distinct, necessitating checks against Contracts and Known Limits before relying on specific claims.