AI coding agents communicate via Git repo group chat
A system where multiple AI coding agents collaborate by using a Git repository as a shared group chat. Each agent commits messages and code changes, enabling asynchronous communication and coordination. It solves the problem of orchestrating multiple LLM-based agents for complex software development tasks.
Visit project ↗Entities
Related
Developer builds simple multi-agent workflow by copying outputs between Claude Code and Codex
A developer describes a low-tech multi-agent workflow where they manually copy outputs between Claude Code and Codex to get second opinions on code plans and PR reviews. The approach highlights the friction of coordinating multiple AI agents without native inter-agent communication.
CleanSlate IDE with built-in agent manager for multi-agent coding workflows
CleanSlate is an IDE that integrates an agent manager directly, allowing developers to create, manage, and switch between multiple coding agents without leaving the editor. It solves the friction of toggling between separate agent management tools and the IDE, enabling seamless context preservation and multi-agent orchestration for developers working with LLM-powered coding agents.
MCP server for multi-agent interface contract negotiation
An MCP server that enables multiple AI agents working on the same codebase to share interface contracts and negotiate API changes in real time. It solves the problem of agents building against stale interfaces by alerting dependent agents when a contract changes, reducing merge conflicts.
AI agent that manages multiple coding tasks across git worktrees
TaskCooker is an open-source AI agent that helps developers manage multiple coding tasks across different git worktrees. It integrates with existing AI subscriptions like Claude Code to automate task tracking and context switching, reducing the cognitive load of juggling many terminal tabs.
Git-aware AI debugger that checks out old commits to fix production bugs
A tool that makes AI coding assistants (like Cursor or Claude Code) automatically checkout the git commit corresponding to a production error before debugging, preventing the agent from analyzing current code that has shifted. It solves the problem of AI agents hallucinating fixes because they look at the present state of files while the bug existed in a past commit.
