agent-channel
NewFile-based named channels for talking between AI coding agents (Claude Code, Codex, OpenCode) across sessions — zero-server, kqueue-backed wake-up.
Summary
This skill enables file-based named channels for communication between AI coding agents like Claude Code, Codex, and OpenCode across different sessions, without needing a server.
- It uses kqueue for efficient wake-up, allowing agents to coordinate tasks, share context, and pass results seamlessly.
Install & Usage
mkdir -p .claude/agentsAdd the configuration to .claude/agents/agent-channel.md
@agent-channelUse Cases
Usage Examples
/agent-channel create review-channel && /agent-channel send review-channel 'Please review the changes in src/main.py'
/agent-channel listen backend-requests --wait 60
/agent-channel send codegen-task 'Generate a REST API for user management' --receiver agent-2
Security Audits
Frequently Asked Questions
What is agent-channel?
This skill enables file-based named channels for communication between AI coding agents like Claude Code, Codex, and OpenCode across different sessions, without needing a server. It uses kqueue for efficient wake-up, allowing agents to coordinate tasks, share context, and pass results seamlessly.
How to install agent-channel?
To install agent-channel: create the agents directory (mkdir -p .claude/agents), then add the config to .claude/agents/agent-channel.md. Finally, @agent-channel in Claude Code.
What is agent-channel best for?
agent-channel is a agent categorized under General. It is designed for: agent. Created by fl4p.
What can I use agent-channel for?
agent-channel is useful for: Coordinate multiple Claude Code agents working on different parts of a large codebase by sending messages through a shared channel.; Pass a complex refactoring task from one agent to another across sessions, preserving context and results.; Enable a frontend agent to request an API endpoint from a backend agent and receive the implementation asynchronously.; Orchestrate a multi-step build pipeline where agents hand off artifacts and status updates via named channels.; Debug a distributed system by having agents log and share findings through a central channel for cross-session analysis.; Automate code review by having a linting agent send issues to a fixing agent through a dedicated channel..