by ahwurm · MCP Server · ★ 42
LocalHarness Run AI agents on the models you already run locally. LocalHarness does not serve models. It sits on top of the one you already serve — vLLM, llama.cpp, Ollama, or LM Studio — and gives it real agents: tools, memory, and permissions, written in YAML instead of Python. Point it at any OpenAI-compatible endpoint and the same agent runs. One main agent reads your task and hands pieces to helpers, each working in its own fresh window with only the tools you allowed it. The bet behind the project: most of what makes an agent good lives in the harness, not the model.
| Stars | 42 |
| Forks | 6 |
| Language | Python |
| Category | MCP Server |
| License | MIT |
| Quality Score | 56.4322183849278/100 |
| Open Issues | 2 |
| Last Updated | 2026-10-05 |
| Created | 2026-05-24 |
| Platforms | browser, claude-code, mcp, python |
| Est. Tokens | ~19k |
These tools work well together with localharness for enhanced workflows:
Explore other popular mcp server tools:
localharness is An open-source, coding agent harness designed specifically for local LLMs. Designed as everything a user needs to jumpstart their own harness (detailed event steams, logs, and performance data) or qu. It is categorized as a MCP Server with 42 GitHub stars.
localharness is primarily written in Python. It covers topics such as agent, agent-framework, agent-harness.
You can find installation instructions and usage details in the localharness GitHub repository at github.com/ahwurm/localharness. The project has 42 stars and 6 forks, indicating an active community.
localharness is released under the MIT license, making it free to use and modify according to the license terms.