Schema | Houtini LM

Houtini LM

Describes the environment variables required to run the server.

Name	Required	Description	Default
`HOUTINI_LM_MODEL`	No	Pin a specific model (or LiteLLM router alias) to use, instead of letting houtini-lm score and choose from the available catalogue.
`HOUTINI_LM_API_KEY`	No	API key for the endpoint, if it requires one (e.g. an OpenAI key like sk-...).
`HOUTINI_LM_THINKING`	No	Control model thinking: 'auto' disables thinking for models detected as supporting the toggle, 'off' forces thinking off on every call, 'on' forces thinking on (useful for bug-hunting).	auto
`HOUTINI_LM_SERIALISE`	No	Whether inference calls are queued one at a time (default for a single GPU). Set to 0 to disable serialisation for OpenAI, a router, or a batching backend like vLLM.	1
`HOUTINI_LM_MIN_TOKENS`	No	Minimum output token budget. Output caps below 4,096 passed by MCP clients are ignored unless this is set to 0.	4096
`HOUTINI_LM_STRUCTURED`	No	Set to 1 so each inference result also carries a structuredContent block (answer, model, tokens, timing, quality flags) for use from your own scripts. Leave off for Claude Code.	0
`HOUTINI_LM_ENDPOINT_URL`	No	The base URL of an OpenAI-compatible endpoint (LM Studio, Ollama, vLLM, SGLang, a LiteLLM router, OpenRouter, OpenAI, DeepSeek, Groq, etc.). Defaults to LM Studio on localhost:1234.	http://localhost:1234

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Server capabilities have not been inspected yet.

Functions exposed to the LLM to take actions

Name	Description
No tools

Interactive templates invoked by user choice

Name	Description
No prompts

Contextual data attached and managed by the client

Name	Description
No resources

This server cannot be deployed

ActivityActive

ResponsivenessNo issues