- Author
- Not stated (published under the yc-software GitHub organization)
- Published
- Undated
- Kind
- GitHub README for an open-source tool
- Original
- github.com/yc-software/qm
- Subject
- Harnesses and agent loops
What it is
The yc-software organization wrote this README for QM, which it says is designed for startups. It says most agents are built as personal assistants. It says those are hard to stretch to a company.
What it says
- QM is a multiplayer agent harness for work, in Slack and on the web.
- Each person and each room gets its own scoped memory, files, permissions, crons and sandbox.
- A headless core with Postgres runs an agent loop called Pi. Pi supports many model providers.
- Orgs pick Strict, Auto (the default) or Dangerous, which has no tool approval gates.
- The agent acts as the person it works for, using their credentials, with everything audited.
- Each deployment runs in the operator's own cloud account.
- The project accepts contributions as human-written text in an adrs folder, not as code.
What evidence it gives
The README gives an architecture diagram, security settings and deployment commands. It offers no benchmark, user study or measured results.
Interests and limits
No individual author is named. Claims about design and security are the project's own and were not tested. Hosted versions are offered through named partners. Only the README was read, not the code or linked docs.