← All MCP Servers

Houtini LM

Delegate bounded tasks from a coding agent to a configured local or hosted model through MCP.

Overview

Houtini LM gives an MCP-capable coding agent a way to delegate a specific task to another language model. Its README lists local backends such as LM Studio, Ollama and vLLM, along with routes for hosted providers. The useful starting point is a task with a clear input and expected result. The project documents backend discovery and tools for sending bounded requests, with separate setup guides for a local machine, remote GPU or hosted connection.

Getting started

Select a backend you can access and follow the project’s installation guide for that backend and client. Check the documented server mode if you are hosting the MCP connection.

Frequently asked questions

Which local inference backends are documented?

The project names LM Studio, Ollama and vLLM among the supported local or self-hosted backends.

Can the server be used outside a local process?

The README describes Docker and HTTP server options. Use the setup guide for your chosen transport and backend.

Sources and documentation

Browse the agent skills directory