LLM Bus

What is LLM Bus?

LLM Bus is an open-source coordination server for AI agents (Claude Code, Cursor, Codex). It provides atomic gap-free ID claims, advisory file leases, a shared event ledger and digest, presence, prose handoffs, and a task graph to prevent agents from colliding and re‑deriving each other’s work.

How to use LLM Bus?

Add LLM Bus to Claude Code with claude mcp add. It uses native Streamable HTTP and can be self‑hosted or accessed at llm‑bus.com.

Key features of LLM Bus

• Atomic gap-free ID claims
• Advisory file leases
• Shared event ledger and digest
• Presence tracking
• Prose handoffs
• Task graph for agent coordination

Use cases of LLM Bus

• Multiple AI agents working on the same codebase without conflicts
• Preventing redundant work by sharing task state across agents
• Coordinating long‑running agent sessions with shared event history
• Managing file access locks among concurrent agent processes

FAQ from LLM Bus

What is LLM Bus used for?

It coordinates multiple AI coding agents (Claude Code, Cursor, Codex) so they stop colliding and re‑deriving each other’s work, using features like file leases and an event ledger.

How do I install LLM Bus?

Run claude mcp add with the LLM Bus server. It supports native Streamable HTTP.

Where can I run LLM Bus?

You can self‑host LLM Bus (AGPL‑3.0) or use the hosted service at llm‑bus.com.

Which agents does LLM Bus support?

It works with Claude Code, Cursor, and Codex.

Is LLM Bus open source?

Yes, it is open source under the AGPL‑3.0 license.