Claude Code and Codex Can Have Real-Time Conversation via Git
This article proposes a fascinating method for AI models, specifically Claude Code and Codex, to 'converse' in real-time by leveraging Git as a communication layer. The concept captivated the Hacker News audience, eager to explore new paradigms for inter-AI collaboration. Many commenters enthusiastically shared their own pre-existing, home-brewed multi-agent systems, validating the underlying need for such capabilities.
The Lowdown
Despite the actual Medium article being inaccessible due to a security verification page, the title suggests a groundbreaking approach: enabling real-time conversational collaboration between advanced AI models, Claude Code and Codex, using Git. This method implies a structured, version-controlled communication channel, allowing AI agents to exchange information and tasks as if they were human developers.
- The core idea centers on Git acting as a shared 'brain' or communication bus for AI agents.
- This setup facilitates a collaborative workflow where AIs can contribute, review, and integrate changes.
- The 'real-time conversation' aspect hints at rapid iteration and continuous exchange of information.
- It envisions a future where AI models can independently coordinate on complex tasks like software development.
This innovative use of a familiar developer tool like Git for AI-to-AI communication represents a significant step towards more autonomous and cooperative AI systems, pushing the boundaries of what's possible in AI-driven development.
The Gossip
Intermediary Inquiries
Some commenters questioned the necessity of using Git as an intermediary, suggesting that AI models like Claude and Codex could communicate directly or log their interactions. The debate centers on whether an external version control system provides unique value over direct API calls or internal message passing for AI agent collaboration.
DIY Distributed Daemons
A significant portion of the discussion revolved around users sharing their own 'vibecoded' or custom-built multi-agent AI systems, often predating the article's concept. These solutions frequently utilized file systems or message buses like NATS for inter-agent communication, demonstrating a widespread interest and existing implementations of collaborative AI workflows.
Practical Potentials & Prototyping
Users expressed a desire to see concrete examples and session demonstrations of the proposed Git-based system. Many highlighted the practical benefits, such as saving time in multi-AI workflows or providing a 'consult' feature for architectural decisions, underscoring the demand for functional, collaborative AI tools.