QM

// Overview

QM is a multiplayer agent harness for work, in Slack and on the web. Y Combinator built it for its own staff and open-sourced it: the project page describes it as letting startups (and YC) work with a fleet of agents, where every employee and project gets one as needed. Each person has an isolated personal workspace, and collaboration happens in shared channels and projects rather than by handing an agent around.

Workspaces come in personal and shared scopes with independent configuration, and identity is unified across Slack and the web interface so the same agent context follows a conversation between them. Administrators set the security posture and choose which models are available, which is what makes it deployable inside a company rather than only on a laptop.

Beyond chat, QM can host custom internal web apps published to a chosen audience, run background automation through crons, watches and webhooks, and search across notes, email, documents and databases. It is written in TypeScript for Node and released under the MIT licence, and it is deployed from your own account onto Fly or AWS.

// What it does

• Personal and shared workspace scopes, each with its own configuration
• One identity across Slack and the web UI, so a conversation carries across surfaces
• Admin-controlled security posture and model selection for the whole organisation
• Custom internal web apps published to a targeted audience
• Background automation via crons, watches and webhooks
• Integrated search across notes, email, documents and databases

// Getting started

QM is scaffolded with its own CLI into a repository you own, then deployed to Fly or AWS. The README suggests handing the deployment itself to a coding agent.

npm exec --yes --package=@yc-software/qm@latest -- \
  qm init . --org <slug> --target <fly-or-aws>
npm install

# hand the repo URL to your coding agent, or follow the deploy docs

qm admin-login

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Give every person in a company their own agent workspace without each one running a separate local setup
• Let humans and agents work in the same Slack channels and projects instead of in private sessions
• Set organisation-wide model and security policy from one admin surface
• Automate recurring internal work with crons, watches and webhooks that agents pick up

// How QM compares

QM alongside other open-source multi-agent systems tools AI/TLDR tracks, ranked by GitHub stars.