qm

YC-Software introduces 'qm', an open-source multiplayer AI agent harness designed for collaborative work in organizations, offering isolated yet shareable agent functionalities across Slack and web interfaces. This new offering has ignited discussions about its strategic positioning against existing agent frameworks like Buzz and Hermes, while also prompting a lively debate over its unique contribution policy: human-written text suggestions only, no AI-generated code. The project highlights the evolving landscape of AI-driven tools and the challenges of managing AI contributions in open-source development.

103

Score

Comments

Highest Rank

on Front Page

First Seen

Jul 31, 6:00 PM

Last Seen

Jul 31, 8:00 PM

Rank Over Time

The Lowdown

QM, developed by yc-software, is an innovative multiplayer AI agent harness engineered for organizational collaboration. It allows individual employees to have isolated AI workspaces while also facilitating seamless teamwork with agents in shared channels, group messages, and projects, whether through Slack or a web interface.

Scoped Workspaces: Each user and room receives a dedicated, isolated environment complete with memory, files, keychains, permissions, crons, web apps, and a durable sandbox.
Model Agnostic: It's built to be open-source friendly, supporting various agent models and harnesses like Pi, OpenCode, Codex, and Claude Code, ensuring flexibility without vendor lock-in.
Key Features: Offers personal and shared scopes, integrated Slack and web access, robust admin controls, custom web app deployment, shareable skills, and background automation capabilities.
Practical Use Cases: Enables collaborative searching of internal knowledge, building and publishing internal applications, automated inbox triage, repository management (testing, PRs), and project tracking.
Security Postures: Implements configurable security postures (Strict, Auto, Dangerous) with pre-declared command policies and auditable actions, mirroring local coding agent security models.
Deployment & Customization: Designed for organization-owned deployments in their own cloud accounts, configurable via a CLI, and supports private forks for extensive customization while maintaining an upstream merge strategy.
Contribution Model: Notably, qm accepts contributions as human-written text descriptions of desired changes rather than code, aiming to maintain design quality and prevent AI-generated 'slop' PRs.

By providing a secure, flexible, and customizable framework for AI agent deployment, qm aims to streamline organizational workflows and empower teams with intelligent automation, all while navigating the complexities of open-source collaboration in an AI-dominated world.

The Gossip

Contributions Conundrum: Human Text Over AI Code

The project's explicit requirement for 'human-written text, not code' for contributions ignited a lively debate. Critics found it ironic for an AI project to disallow AI-generated code, suggesting a 'psychosis problem' in the industry. Supporters, including maintainers, argued it's a practical and sensible approach to avoid 'slop PRs,' ensure design quality, and prevent context loss from AI-generated boilerplate, preferring well-thought-out specifications over poorly coded submissions.

Agent Arena: Comparing QM to Competitors

Commenters actively compared `qm` to other 'openclaw-like' agent systems such as Hermes, Buzz, and Dirge, discussing their features, complexity, and suitability for various tasks. Many shared their practical experiences, from using agents for monitoring and fixing CI failures, creating RCAs, optimizing database queries, and generating charts, to triaging emails and managing RSS feeds. A key takeaway was the challenge of scoping and the need for customizable, rather than feature-packed, solutions.

YC's Venture: Internal Tool or Strategic Launch?

The discussion also included speculation about `qm`'s origin and strategic purpose. Some wondered about 'yc-software' itself, while others posited that `qm` might be an internal YC tool 'rushed out the door' to compete with similar projects like Buzz, or even that YC was becoming a startup itself by shipping software. This highlights Hacker News's keen interest in the strategic motivations behind YC's public actions.