Orca MCP Server

Lets an assistant turn natural-language descriptions into finite state machines and verify or compile them, catching topology bugs before code runs.

What you can do

Orca MCP Server exposes tools for working with finite state machines: parse_machine, verify_machine and compile_machine handle parsing, verification and compilation, while generate_machine, generate_multi_machine, refine_machine and generate_actions (with use_llm) use an LLM to produce or refine machine definitions from natural language. The stated goal is to catch topology bugs before code runs.

When to use it

Useful when you want an assistant to draft, refine or check state-machine models from a description, and to validate their structure before generating code. Best suited to design and verification work rather than runtime execution.

Requirements

Runs locally as a stdio process, started with npx from the npm package @orcalang/orca-mcp-server, so Node.js and network access for the package download are needed. LLM-backed tools require the ANTHROPIC_API_KEY secret; ORCA_MODEL and ORCA_PROVIDER select the model and provider (anthropic, openai, grok or ollama). The parse, verify and compile tools do not require the API key.