Orchex

Orchex is a local MCP server that lets an assistant orchestrate work across six LLMs, planning, parallelizing, self-healing and routing tasks automatically.

What you can do

Plan and break down tasks automatically before execution, according to the server's description of autopilot orchestration. Run work in parallel across multiple LLMs rather than one model at a time. Route each task to one of six LLMs. Recover from failures automatically (self-heal) during a run. No individual tools are listed in the manifest, so exact operations are not documented.

When to use it

Worth considering when you want an assistant to coordinate multi-step work across several model providers instead of relying on a single model, and you already hold API keys for those providers.

Requirements

Runs as a local process via the npm package @wundam/orchex over stdio, so Node.js is needed. It reads provider API keys from the shell environment: ANTHROPIC_API_KEY, DEEPSEEK_API_KEY, GEMINI_API_KEY and OPENAI_API_KEY. An optional ORCHEX_API_URL overrides the default cloud API endpoint. Desktop only.