Agent Harness Docs
Start with installation and the first-run flow, then use the CLI, API, configuration, examples, troubleshooting, and how-to guides as your working reference.
Introduction
What Agent Harness is for and how to think about the product.
Install
Install Agent Harness from a source checkout or Docker Compose.
CLI Reference
The public Agent Harness command surface and safe first commands.
Quick Start
The intended works-in-10-minutes path for a new Agent Harness user.
SDK and MCP API
The supported Agent Harness integration surfaces for agents and automation.
Configuration
The current config shape and the fields most operators need first.
Examples
The supported example configs and what each one demonstrates.
How-To Guides
Practical recipes for common Agent Harness workflows.
FAQ
Answers to common installation, privacy, model, and workflow questions.
Model Intake and Benchmarking
Safely discover local models, benchmark supported task suites, and review results without exposing private evaluation designs.
Operations
The operator workflow for approvals, failures, retries, and recovery.
Pro Tips
Practical habits for safer, clearer Agent Harness runs.
Compatibility
What soft-stable means for the Agent Harness CLI, config, and artifacts.
Roadmap
The current roadmap for Agent Harness — what shipped, what is in progress, and what comes next.