OpenHarness

Why we included this project

OpenHarness is a Python codebase that wires the pieces around an LLM into a working agent: tool use, skills, memory, and multi-agent coordination, all reachable through a single `oh` command. It also ships ohmo, a personal agent that lives in Feishu, Slack, Telegram, or Discord and can fork branches, write code, run tests, and open pull requests on its own, running on your existing Claude Code or Codex subscription with no extra API key. So it works on two levels: as a reference for researchers and builders who want to understand or extend agent internals, and as a ready-to-run assistant for developers who want an agent that keeps working for them over long sessions. If you plan to let an autonomous agent loose, the permission modes, hooks, and path-level rules are worth reading for keeping it contained.