qm – Multiplayer agent harness for work
Points and comments are a snapshot, not live.
QM is a multi-agent harness for startups, offering scoped workspaces in Slack and on the web.
QM is an open-source agent harness designed for startups, giving each employee an isolated workspace with scoped memory, files, keychain, permissions, crons, web apps, and sandboxes. It integrates with Slack and the web, supports multiple models (Pi, OpenCode, Codex, Claude Code), and offers admin controls, shared skills, background work via crons, and posture-based security (Strict, Auto, Dangerous). Deployed via a CLI tool in the user's own cloud account under MIT license.
What commenters are saying
Commenters largely supported the project's unusual contribution policy asking for human-written text instead of code PRs, comparing it to SQLite's approach. Several maintainers noted it reduces low-effort submissions and allows them to implement features using their own AI harnesses. Some joked that the policy effectively asks users to "write our prompts for us." A secondary discussion explored different open-source agent harnesses, with users sharing preferences for Hermes, Dirge, and Cecli.