Physics MCP Server

--- title: NLI Tool kind: reference header_svg: src: "/assets/svg/tool-nli-hero.svg" static: "/assets/svg/tool-nli-hero-static.svg" title: "NLI Tool" animate: true theme_variant: "auto" reduced_motion: "auto" --- {% assign header_svg = page.header_svg %} {% include header-svg.html %} # NLI Tool [Home](../../README.md) · [Architecture](../Architecture.md) · [Configuration](../Configuration.md) · Tools: [CAS](CAS.md) · [Plot](Plot.md) · [NLI](NLI.md) Natural Language Interface parses free-text physics requests into structured tool calls. Tool - `nli_parse` - Params: `text` (string) - Returns: `{ intent, args, confidence?, explanation? }` - Example request: ```json {"jsonrpc":"2.0","id":"1","method":"nli_parse","params":{ "text":"Plot y = x^2 from -5 to 5" }} ``` Intents - CAS: `cas_evaluate`, `cas_diff`, `cas_integrate`, `cas_solve_equation`, `cas_solve_ode` - Plot: `plot_function_2d`, `plot_parametric_2d`, `plot_field_2d` - Unknown: `unknown` with an explanation when parsing fails Operation - Primary (optional): Calls a local LM API (`/chat/completions`) with a physics-aware system prompt when configured. - Fallback (default when no LM configured): Uses a built-in rule-based parser with curated physics patterns for common tasks. This requires no external services and works entirely offline. Configuration - Optional: LM Studio (or any OpenAI-compatible local endpoint). Not required for core calculations. - Environment (optional): - `LM_BASE_URL`: Base URL to your OpenAI-compatible endpoint (e.g., `http://localhost:1234/v1`). If not set, the NLI automatically uses the rule-based fallback. - `DEFAULT_MODEL`: Model name to use when `LM_BASE_URL` is set. - `LM_API_KEY`: API key if your local server requires it. - `NLI_DISABLE_LM`: Set to `true` to force the rule-based parser even if `LM_BASE_URL` is configured. - `NLI_LM_TIMEOUT_MS`: Milliseconds to wait for the LM before falling back (default: 2500ms). Why a local LM helps (optional) - Lower latency and fewer retries on complex requests → faster end-to-end results. - Uses your GPU (when available) to accelerate parsing. - Privacy and cost benefits from keeping tokens local. Quick setup (LM Studio) 1. Install and run LM Studio. 2. Set `LM_BASE_URL` to the OpenAI-compatible endpoint (e.g., `http://localhost:1234/v1`). 3. Set `DEFAULT_MODEL` to your chosen local model. Schemas & Prompt - Types and prompt: `packages/tools-nli/src/schema.ts` - Logic and LM usage: `packages/tools-nli/src/index.ts` Notes - If `LM_BASE_URL` is not provided (as in the default Windsurf MCP config), the NLI will gracefully and immediately use the rule-based parser. This keeps the system independent of LM Studio by default while still allowing you to opt-in to LM-powered parsing later. A gentle quip: we keep the parsing Hamiltonian simple-no unnecessary terms, just enough to reach the ground truth.