Agent Bus: Local message bus for AI agents without daemon or cloud

Agent Bus: Local Message Bus for AI Agents

Agent Bus is a lightweight, local message bus designed for AI agent development. It facilitates communication between multiple AI agents by leveraging a single SQLite file for message persistence. This approach eliminates the need for external daemons or cloud infrastructure, simplifying deployment and reducing operational overhead. Agent Bus is specifically optimized for use with models like Claude Code and Codex, making it a valuable tool for developers building complex AI systems that require efficient inter-agent communication.

What it Does

Agent Bus acts as a central hub for AI agents to exchange messages. Agents can publish messages to specific topics, and other agents subscribed to those topics will receive the messages. The entire message queue and state are stored within a single SQLite database file. This means that all communication is local to your development environment or server, ensuring data privacy and control. The absence of a separate daemon process simplifies setup and maintenance, as there's only one core component to manage.

Key Features

• Local Operation: Runs entirely on your local machine or server, no external dependencies or cloud services required.
• SQLite Persistence: Utilizes a single SQLite file for storing all messages, ensuring data durability and easy backup.
• No Daemon Required: Simplifies deployment and management by eliminating the need for a separate background process.
• Optimized for Code Models: Specifically tuned for use with AI models like Claude Code and Codex, enhancing performance for code-related agent tasks.
• Topic-Based Messaging: Supports a publish-subscribe model for efficient message routing between agents.

Who it's For

Agent Bus is an ideal solution for AI developers who are building multi-agent systems and prioritize a simple, self-contained architecture. This includes:

• AI Researchers: Experimenting with agent interactions and communication patterns without complex infrastructure setup.
• Software Engineers: Integrating AI agents into existing applications where a local, low-latency communication layer is beneficial.
• Hobbyists and Makers: Developing AI projects that require reliable agent communication without the cost or complexity of cloud services.
• Developers using Claude Code or Codex: Leveraging the optimizations for these specific models to enhance the performance of their code-generating or code-understanding agents.