Skip to content
Automation
MCP

fastskills

Ships with onemancompany: Build Your Agent Company with OMC

From plugin
onemancompany
45817 skills1 MCP

Claude Code

$ npx @1mancompany/onemancompany

Security report

Found

Low risk

  • Known flaws

    None known

  • Signed build

    No

    No build provenance published

  • Maintained

    Aug 26, 2026

    Last push

  • Licence

    Apache-2.0

  • Runs as

    uvx

    On your machine

  • Deprecated

    No

Signals from public registries, not an audit. Checked Nov 12, 2026.

onemancompany README

OneManCompany


CI

npm version

GitHub stars

License

Last commit

Discord

HuggingFace Paper

OneManCompany — Pixel Art Office

Homepage  ·  中文文档  ·  Docs  ·  Talent Market  ·  Issues

Others use AI to write code. You use AI to run a company.

Linux runs servers. OneManCompany runs companies.

Not building a company — building any company.

OneManCompany is an open-source OS that lets anyone build and run a complete AI-powered company from their browser. It doesn't replace tools like Claude Code, Codex, or other agent frameworks — it operates one level above them as an agentic operating system, orchestrating hierarchical AI teams built on top of these underlying tools.

You are the CEO — the only human. Everyone else — HR, COO, engineers, designers — are AI employees that think, collaborate, and deliver real work autonomously. (No slacking, no sick days, no raise requests — just the occasional need for a pep talk.)

⚙️ An OS, not just a product: Swap the plan, swap the clan — brand new company in your hand. One runtime abstracts away AI differences. Skills install like phone apps. Same system, different direction and team — entirely new company.

🤖 Real agents, but top talent welcome: Built-in execution engine, task scheduling, retry & fault tolerance, multi-agent communication protocols. OpenClaw, Claude Code, and other powerful agents can move right in as employees. Yes, there's a visualization — but that's just the surface. Underneath runs a full operating system. The pixel-art office is just the tip of the iceberg.

🏢 Real company, not a chatbot: Interview, review, promote the few — fire the ones who haven't got a clue. Yes, your AI employees get nervous. Org chart, hiring pipelines, performance reviews, approval chains, meeting rooms — modeled after Fortune 500 management.

🤝 Agents that actually collaborate: Your team talks to each other — not just to you. Engineers pull in designers, COO coordinates the big picture, employees call their own meetings to align — multi-agent meetings, cross-functional delegation, structured handoffs. Real teamwork, not parallel chat windows pretending to be a company.

🛡️ No more hallucination roulette: Download counts and real reviews don't lie — trust the community. Tired of AI that talks big and delivers nothing? We ship with a Talent Market of battle-tested, community-scored AI employees. No self-proclaimed "I'm an expert in everything" agents allowed.

📈 Gets smarter with use: Your team forgets nothing — even the AI employees it forgets. Every retrospective and 1-on-1 becomes lasting experience. Tell them once, they learn. Tell them twice, they remember. Still no good? Replace them — the next one already heard what happened to the last one.

npx @1mancompany/onemancompany

One command. A full AI company in your browser. No Python, no Docker, no config — just run it.

If this project is useful to you, please consider giving it a ⭐ — it helps others discover OneManCompany.


Features

OneManCompany — Task Tree

Hierarchical task breakdown — CEO delegates, AI employees execute.

Talent Market — Community-Verified AI Employees

Talent Market — find, hire, and onboard community-verified AI employees.


Demos

Make me a game

"Make me a game" CEO gives one sentence — the team builds a complete game autonomously.


How It Works

You open a browser. You see a pixel-art office. Your AI employees are at their desks, pretending to look busy.

You type: "Build a puzzle game for mobile"

  1. Your EA receives the task and routes it
  2. Your COO breaks it down and dispatches subtasks
  3. Engineers, designers, and QA work autonomously
  4. They hold meetings to align when needed
  5. Work goes through review, iteration, and quality gates
  6. You get notified and approve the final result

You manage. AI executes.

CEO (You, the only human who gets coffee breaks)
  └── EA ── routes tasks, quality gate
        ├── HR ── hiring, performance reviews, promotions
        ├── COO ── operations, task dispatch, acceptance
        │    ├── Engineer (AI)  ← hired from Talent Market
        │    ├── Designer (AI)  ← hired from Talent Market
        │    └── QA (AI)        ← hired from Talent Market
        └── CSO ── sales, client relations

Founding team (EA, HR, COO, CSO) comes built-in. Need more people? HR searches the Talent Market — a community-verified marketplace of AI employees.


What You Can Build

AI CompanyWhat It Delivers
🎮 AI Game StudioProduction-grade games with full playtesting and iteration cycles
📖 AI Manga StudioSerialized comic stories with consistent art and narrative
💻 AI Dev AgencyShip software products end-to-end
🎨 AI Content StudioMarketing campaigns, branded content, and media production
🔬 AI Research LabLiterature review, data analysis, and report generation

These aren't toy demos — each AI company produces product-level deliverables through a full team of collaborating AI agents.

How We're Different

Typical Agent OrchestratorsOneManCompany
Agent systemWrapper / visualization on third-party agentsOwn agent runtime — execution engine, retry, communication, all built from scratch
Agent collaborationParallel execution, no real interactionMulti-agent meetings, cross-functional delegation, structured handoffs
Where do agents come from?You find and configure them yourselfC-suite built-in on Day 1 + HR hires from a verified Talent Market
Management methodologyAd-hoc, improvised by LLMSystematic — modeled after Fortune 500 org structures and management science
Agent architectureFlat task runners, BYOAVessel + Talent separation — modular architecture with Harness protocols
Execution modelHeartbeat polling / loopEvent-driven, zero-idle, on-demand dispatch
OrganizationSimple task queuesFull company simulation — org chart, reviews, coaching, meetings
DeliverablesSingle-point task outputsMulti-iteration project delivery with quality gates

Quick Start

You only need Node.js 18+ and Git. Everything else is installed automatically.

npx @1mancompany/onemancompany

macOS

@1mancompany
Ships withonemancompany
45817 skills1 MCP
Get the whole plugin

Stars on GitHub

Momentum

stars in total
458
Star history of 1mancompany/OneManCompany on GitHub, a chart by star-history.com

More MCP servers from other plugins.

playwright

playwright

MCP

Drives a real browser so the agent can open pages, click, fill forms and read what actually rendered.

windmill12d18k
@windmill-labs@windmill-labsInstall MCP
pinchtab

pinchtab

MCP

Ships with pinchtab: High-performance browser automation bridge and multi-instance orchestrator with advanced stealth injection…

@pinchtab@pinchtabInstall MCP
open-computer-use

open-computer-use

MCP

Ships with open-computer-use: open-computer-use is an open-source Computer Use service wrapped as MCP. Any AI agent or MCP client…

@ifuryst@ifurystInstall MCP
atlas

atlas

MCP

Ships with babysitter: Enforce obedience on agentic workforces. Manage extremely complex workflows through deterministic,…

babysitter10d1.9k
@a5c-ai@a5c-aiInstall MCP

Questions about fastskills.

Which coding agents can use fastskills?
fastskills works in any agent that reads an MCP config file: Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and Windsurf. Each agent reads it from its own file; this page names the file for each one.
Is fastskills safe to install?
Low risk. Signals from public registries, not an audit.