Peers — let AI coding sessions talk to each other
Running several AI coding sessions in parallel usually means they are blind to each other — duplicating work and drifting apart. Peers is a local coordination server that connects multiple Claude Code and Codex sessions so they can discover each other, exchange messages, request reviews, share artifacts and collaborate through shared scratchpads. Each session connects via a lightweight MCP server over stdio to a local ASP.NET Core broker backed by SQLite — everything stays on localhost, no cloud required. Open source: gitlab.com/Dave3991/peers-mcp
Project information
- Technology used: .NET 10, ASP.NET Core, SQLite, Model Context Protocol (MCP)
- Project goal: Turn isolated AI agents into a coordinated team
- Date: 2026
- Repository: gitlab.com/Dave3991/peers-mcp