oh-my-agent
A multi-agent harness that verifies results by artifacts: stop hooks, an independent judge and an event log for Claude Code, Codex, Cursor and more
Install
bunx oh-my-agent@latestRequires bun, uv and serena.
This is third-party code. Review the repository files before installing.
What it does
oh-my-agent keeps skills, workflows and rules in one .agents folder and projects them into each supported agent's native layout. A stop hook refuses to end the session until the project's typecheck, test or lint scripts pass, and a gate command checks whether a workflow left the required artifacts. An independent judge with fresh context re-verifies every criterion each round, and decisions go to an append-only log. It ships role agents for architecture, backend, frontend, databases, debugging, QA, DevOps and more, with ready presets.
Who it is for. For developers and teams who need factual verification rather than an agent's claim of done.
Good fit when
- The agent often claims tests pass when they do not
- You need the same skills and rules across Claude Code, Codex and Cursor
- You want to split work among role agents using a preset
Not a fit when
- The project has no check scripts to rely on
- Extra tools such as bun, uv and serena cannot be installed
Example request
Implement the signup endpoint and do not finish until all tests and the linter passLimitations
The install script installs bun, uv and serena if missing. Gates are only useful when the project has working typecheck, test and lint scripts. Do not mix APM and bunx installs in one project, they drift apart.
How to disable. Delete the .agents folder and the files it generated in .claude, .cursor, .codex and other agent folders, plus the hooks in their settings.
Security check
- The install script is piped from the network and installs third-party tools
- Hooks run project scripts and control agent sessions
README in short
The README opens with agents reporting success without proof and describes stop hooks, gate commands, the judge and the event log. Install via script for macOS, Linux and Windows, via bunx, or skills only via APM. It includes a preset table, supported agents and role agent descriptions. The README is translated into many languages, including Russian. MIT licensed.
FAQ
Can I install only the skills?
Yes, via Agent Package Manager: apm install first-fluke/oh-my-agent. Workflows, hooks and the CLI are not installed that way.
Which agents are supported?
Claude Code, Codex CLI, Cursor, OpenCode, Amp, GitHub Copilot, Qwen Code, Kiro CLI and others.
Related
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Anthropic's coding agent for the terminal, IDE and GitHub: understands your codebase, runs tasks and handles git