MCO: multi-agent tool for Hermes Agent

CLI orchestration that runs AI coding agents in parallel and compares their raw answers

What MCO does

MCO runs the agents and models you choose side by side for code review, implementation, architecture analysis or CI checks. You select the team explicitly with --providers or --agent, and it never guesses a team from the binaries it finds. The mco review command is read-only, while mco run with --execution-mode write gives agents workspace write access. You can preview a run with --dry-run, stream progress live or as JSONL, or chain agents so one builds on the previous result.

Its main design choice is that answer text stays opaque. MCO keeps each invocation's full raw output and operational status, and it does not convert prose into findings, severity, confidence or an automatic consensus. Built-in providers cover Claude Code, Codex CLI, Gemini CLI, OpenCode, Qwen Code, GitHub Copilot CLI, Hermes, Pi, Grok Build, Cursor CLI and Antigravity CLI. Each provider CLI handles its own installation, authentication and sandboxing.

Key features

Parallel runs across selected agents and models
Read-only review mode and a write mode for implementation tasks
Dry-run, live stream and JSONL event output
Chained runs and divided scope with --divide files or --divide dimensions
Raw answers kept without an automatic consensus
mco doctor --json to check which agent CLIs are available

When to use it

Get a multi-agent review of a repository before merging
Compare how different models implement the same change
Run a read-only multi-agent check in CI
Call MCO from inside another coding agent to get second opinions

Who it is for: Developers who already use several AI coding CLIs and want one command to run them together and compare the results.

How it fits with Hermes Agent

Hermes is a built-in provider with the CLI name hermes and provider ID hermes, so Hermes Agent can sit in a review or implementation team next to other agents. The repository is also tagged hermes-agent.

How to install MCO

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

npx @tt-a1i/mco@latest install
mco doctor --json

Requirements: The agent CLIs you select must be installed and authenticated on their own

FAQ

What is MCO?

MCO is a lightweight, CLI-first layer that runs selected AI coding agents in parallel on one task. It keeps each agent's raw answer and status so you can compare them before acting.

Does MCO work with Hermes Agent?

Yes. Hermes is one of the built-in providers, with the CLI name hermes and provider ID hermes. You include it in the team with the --providers option.

How do I install MCO?

Run npx @tt-a1i/mco@latest install, which installs the CLI and its bundled mco-cli Skill. Then run mco doctor --json to see which agent CLIs it can find.