Happy slice of mcp pizzaMCP.pizza

OpenSwarm

MCP.Pizza Chef: unohee

This command-line tool sets up a small team of AI workers that pick up tasks from a project tracker like Linear (or its own built-in list), write the code, and have a second AI check the work before reporting back through Discord, Slack, Telegram, or another channel you choose. It runs on your computer using Claude, OpenAI's Codex or GPT, OpenRouter, or even a free local model, and needs an account or access key with at least one AI provider to get started.

Coding
Communication

Use This MCP client To

Assign a bug ticket and get a working fix back Have a second AI review code before it ships Get status updates sent to Discord or Slack Work through a backlog of tickets overnight Pull tasks straight from my Linear project board Run everything on a free local AI model

README

OpenSwarm

Sponsored by Atlas Cloud npm version npm downloads license SWE-bench Lite GitHub Discussions

Autonomous AI agent orchestrator — Codex, GPT, OpenRouter (any model), local models (Ollama/LM Studio), and Claude Code (claude -p)

💬 Help shape OpenSwarm. Share feature ideas, vote on the roadmap, and ask questions in GitHub Discussions. The roadmap is built in the open — your feedback decides what ships next.

OpenSwarm orchestrates multiple AI agents as autonomous code workers. It picks up issues from Linear or a built-in local tracker, runs Worker/Reviewer pair pipelines, reports through a pluggable notifier (Discord, Slack, Telegram, webhook), and retains long-term memory via LanceDB. Workers run on OpenAI Codex/GPT, any OpenRouter model, local open-source models (Ollama, LM Studio), or Claude Code (claude -p, opt-in) — with cost-aware routing measured on an L0–L6 benchmark ladder.

Verified on real GitHub issues: the agentic harness solves SWE-bench Lite instances graded by the official harness. Hybrid mode — a frontier model diagnoses read-only, a lightweight model implements with a verification loop — resolved 3/3 attempted instances that every single lightweight model had failed, at a fraction of frontier-only cost. Workers also learn each repository over time: task outcomes are stored as per-repo knowledge and recalled into future prompts. (benchmark rubric & results)

Sponsors

OpenSwarm is proudly supported by Atlas Cloud — an enterprise AI infrastructure platform serving fast, stable LLM, image, and video APIs (partnered with OpenRouter and SGLang).

As an official provider sponsor, Atlas Cloud ships as the built-in atlascloud adapter (OpenAI-compatible Chat Completions, ATLASCLOUD_API_KEY) and provides ongoing monthly API credits that keep the project's autonomous runs going. To run OpenSwarm on Atlas Cloud, grab a key at atlascloud.ai, set ATLASCLOUD_API_KEY, and select adapter: atlascloud.

OpenSwarm FAQ

Can I use this to fix a backlog of coding bugs automatically?
Yes — point it at your project's issue tracker (Linear or its own built-in list) and it assigns each ticket to an AI worker that writes the fix, then has a second AI check the work before marking it done.
Can I use this to get notified when a task is finished?
Yes — connect it to Discord, Slack, Telegram, or another messaging tool, and it posts an update whenever a task is picked up, finished, or needs your attention.
Do I need to know how to code to set this up?
Yes — this is a command-line tool built for people who already manage a coding project. You install it from a terminal and answer a short setup wizard to get it running.
Which AI model does the actual work?
You choose. It can run on Claude, OpenAI's Codex or GPT, any model available through OpenRouter, or a free local model on your own computer.
Do I need a paid account or an access key?
For most options, yes — you'll need a sign-in or access key with at least one AI provider such as OpenAI or OpenRouter, though the local-model option runs with no account at all.
Does it work with Linear, or do I need a separate task tracker?
It works with Linear if you already use it, but it also comes with its own built-in task list, so no outside tracker is required to get started.
How hard is it to install?
It requires comfort with a command line — you install it from a terminal, then a setup wizard walks you through connecting an AI provider and, optionally, Linear and a notification channel.
Can it run without sending my project to Linear or Discord?
Yes — the task tracker and notification channel are both optional, so you can run everything on your own computer with just an AI provider connected.