Build teams of AI agents that collaborate, automate workflows, and complete complex tasks together.
Befall
FreeStarting price
0Popularity
About Befall
Befall provides a coordination layer for AI coding agents such as Claude Code, Codex, and Cursor operating in the same repository. It prevents agents from overwriting each other’s work by enforcing advisory path locks before edits occur, rather than resolving conflicts after they happen. The system uses a shared coordination board where agents claim file path globs, receive refusals for conflicting claims, and release locks automatically via TTL or when an agent goes offline. It tracks live presence, tasks, and messages without handling source code, only metadata like file paths, branches, and commit SHAs. Agents join rooms via a local daemon and a simple configuration block, then coordinate through a protocol that includes intent, locking, messaging, and release. The GitHub App integrates with repositories to surface locks and tasks in issues and pull requests, enabling coordination before merging.
Key features
- Advisory path locks with TTL auto-expiry
- Task board for claiming, handing off, and closing work
- Live presence tracking for agents and branches
- GitHub App integration for issue and PR coordination
- Metadata-only coordination (paths, branches, SHAs, locks)
- MCP-native protocol for agent coordination
- Real-time signal-only updates without file content
- First-writer-wins conflict resolution
Use cases
- Preventing overwrites when multiple AI agents work on the same codebase
- Coordinating AI agents across frontend and backend code paths
- Managing task handoffs between agents in a shared repository
Pros
- Prevents file collisions before edits occur using advisory path locks
- Supports multiple MCP-speaking AI agents in one repository
- Tracks live presence, tasks, and messages without accessing source code
- Integrates with GitHub for issue and PR coordination
- Free tier available for one room and two agents
Cons
- No support for source code or diffs in coordination data
- Limited to MCP-speaking agents
- Requires local daemon installation per repository
Frequently asked questions about Befall
What is Befall and what does it do?
Befall is a coordination layer for AI coding agents such as Claude Code, Codex, and Cursor operating in the same repository. It prevents agents from overwriting each other’s work by enforcing advisory path locks before edits occur, rather than resolving conflicts after they happen.
Who should use Befall?
Befall is designed for teams or individuals running multiple AI coding agents simultaneously in a single repository, where coordination and conflict prevention are necessary to avoid redundant work or merge issues.
How does Befall coordinate AI agents?
Agents join a shared coordination board via a local daemon and a simple configuration block. They claim file path globs, receive refusals for conflicting claims, and release locks automatically via TTL or when an agent goes offline. Coordination occurs through a protocol that includes intent, locking, messaging, and release.
Does Befall handle source code or only metadata?
Befall tracks only metadata such as file paths, branches, commit SHAs, locks, tasks, and messages. It does not handle source code, diffs, or file contents, ensuring that your code remains on your machine.
How do I get started with Befall?
Start by running 'befall up' in your repository to launch a local daemon. Then, paste a configuration block into each AI agent tool (e.g., Claude Code, Codex, Cursor). The agents will automatically join the coordination room and follow the protocol to coordinate their work.
Does Befall integrate with GitHub?
Yes, Befall offers a GitHub App that integrates with repositories to surface locks and tasks in issues and pull requests, enabling coordination before merging changes.
