MCP server that delegates Claude Code agent work to local models (llama.cpp / vLLM) on your own hardware
Hand Claude Code's grunt work to a model running on your own hardware.
localagents is an MCP server that gives Claude Code a runagent tool. Each call starts a full headless Claude Code session — same tools, same CLAUDE.md, same working tree — except its API traffic goes to a llama.cpp or vLLM server you run instead of to Anthropic. Claude writes the brief, the local model does the work, Claude reviews the result. Your Anthropic token budget goes on the parts that need it.
A 27B Qwen on one GPU is perfectly capable of "add a CLI for this module and tests to match"; Opus is better spent on the design conversation than on watching pytest run. Two local agents in parallel can build the two halves of a package against a pinned interface.
Status: early. It works, I use it daily, and the interface will move. It targets llama.cpp and vLLM specifically; Ollama isn't a goal.
Three things make this more than an environment variable:
1. A registry that is probed live. models.yaml lists where servers are and a menu of model names. What each server is actually serving right now, its real context window, and how many slots are busy are discovered on every call. You bring models up and down by hand — the server never launches anything — and when Claude needs a model that isn't running it asks you for it by name.
2. A shim between Claude Code and the backend. Claude Code sends things local chat templates reject, and local servers fail in ways Claude Code doesn't recognise. The shim fixes both directions (details below) and writes a requests.jsonl per job so you can see exactly what we
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/ccebelenski--localagents.html)
🦆🧠 Give your small robot a brain. Tell a Microduck, a Reachy Mini, a LeRobot arm or any ROS base what you want in plain language. An LLM (Claude, OpenAI, Gemini, Grok, or a local model via Ollama, vLL
lawrence3699/jev-styleSmall, calibrated decision models on your own machine: systemone-compatible local server, 6 agent skills, Claude Code guard, MCP tools. Weights on Hugging Face.
1102tools-dev/federal-contracting-agentsFive packaged federal acquisition agent previews for Codex and Claude Code.
King4s/Jev-AI-SkillOne AI skill + MCP server for Claude Code, Codex and Hermes: Jev (TypeSafe) gates large-model turns (event triage, review verdicts, owner questions, tool choice), runs build-and-repair loops with inde
janwilmake/agent-codemodeLet the scripts your coding agent writes call your connected MCP servers (no extra auth)
CogniSwitch/KL4ATurn SOP, policy and regulatory documents into human-verified Open Knowledge Format (OKF) bundles — every claim carries its source span. Desktop app + CLI + MCP server.
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.