Automatic Control for intelligence in action.

Quickstart   ·   Documentation   ·   How it works

Tell an agent what to do in your own words. A language model plans and writes short programs; The Harness runs them in an environment the agent can see and act in. Watch, send a message, or take over at any time.

The agent thinks while it acts: between model calls, its programs, skills and reactive rules keep working. What it learns along the way—notes, tested skills and rules—applies in the same run.

Getting started

npm install -g @lmmslab/harness
harness

The Harness needs Python 3.11 or newer. You can also install it with pip install the-harness.

harness opens a conversation in your terminal. The first time, it asks which model to use and for its key; then type a task. Any model works: hosted ones through OpenRouter, OpenAI, Anthropic and other providers, or a model you run yourself.

See Install and the Quickstart for a real task on your own computer.

Commands

Use The Harness in the terminal, run one task from the CLI, drive it from another program with JSON lines, call it from apps, editors and other agents over MCP and ACP, or build on the Python API.

harness                 # talk with the agent in the terminal
harness run "task" -p   # do one task and print the answer (--json: every event)
harness run rpc         # drive a conversation with JSON lines, from another program
harness install         # packages (also: list, remove)
harness library         # what the agent learned
harness replay          # a run again in the terminal
harness doctor          # what this computer has and lacks

Where it works

pip/npm install only the small core. Environments come in packages, installed by themselves the first time a task needs them.

Packages

Documentation

• Quickstart
• Automatic Control: the ideas behind The Harness
• Models and providers
• Interrupt and take over
• Let it learn an environment
• Write a skill, write a package, add an environment

Development

pip install -e .
harness install ./packages/computer ./packages/browser ./packages/workspace --link --yes
harness doctor

See CONTRIBUTING.md and the validation guide.

Citation

@misc{lmmslab2026harness,
  author = {{LMMs-Lab}},
  title = {The Harness: Automatic Control for Intelligence in Action.},
  howpublished = {\url{https://github.com/EvolvingLMMs-Lab/the-harness}},
  year = {2026}
}

License

MIT; third-party material retains its own terms. See LICENSE and CONTRIBUTORS.md.