# AgentRig

> **Note:** This is a summary overview using the LLMs.txt format (https://llmstxt.org/). Each section links to its full documentation file in plain text format.

A meta-harness for agent harnesses — install best-practice agent rules, skills, and surfaces into any repo with one command.

- [Agent surfaces](./llms/agent-surfaces.txt): Which files AgentRig projects into which agent ecosystem, and why.
- [Commands](./llms/commands.txt): Every AgentRig CLI command and flag.
- [Evaluating the harness](./llms/evals.txt): Three layers — install completeness, quality probes, fixture-based dynamic eval with paired sign-test lift — plus judge calibration. Honest about what each does and does not prove.
- [Getting started](./llms/getting-started.txt): Install AgentRig and scaffold a best-practice agent harness in any repo in under a minute.
- [AgentRig](./llms/index.txt): A meta-harness for agent harnesses. Investigate any repo, install best-practice agent rules, skills, and surfaces — every agent reads from one source.
- [Principles](./llms/principles.txt): The 12 principles AgentRig encodes — the contract a healthy agent harness should satisfy.