Health + schema is the missing layer… I went one step lower and focused on coordination primitives (claiming, locks, drift checks) exposed via MCP.

MCP solves the wiring problem nicely, but once you have more than one coding agent running, coordination becomes the real bottleneck.

I ran into repeated issues with agents editing the same files, losing decisions, and drifting away from the spec. I ended up building a small Rust MCP server that adds task claiming, file-level locking, and drift checks against SPEC.md.

It’s intentionally boring infrastructure… single binary, file-based state, deterministic orchestration. The AI tools stay interchangeable.

Repo if anyone’s curious: https://github.com/nxtg-ai/forge-orchestrator

Forge – A 3 MB Rust binary that coordinates multiple AI coding agents via MCP https://github.com/nxtg-ai/forge-orchestrator

Forge is an orchestration layer that sits between AI coding tools and your codebase. It's a single Rust binary (~3 MB, zero runtime deps) that runs as an MCP server over stdio. Any MCP-compatible AI tool can call it.

MIT licensed. Whitepaper with the full architecture: https://nxtg.ai/insights/forge-whitepaper