Run Your AI Agents Like a Company with Paperclip

Ten agent terminals open, no idea which one is doing what, and a reboot wipes the board. Paperclip — a 102k-star open-source orchestration app — replaces that chaos with one dashboard where agents get goals, roles, and budgets.

Why This Skill Matters

Paperclip is a Node.js server and React UI that orchestrates a team of AI agents to run a business. The README puts it plainly: if OpenClaw is an employee, Paperclip is the company. You bring your own agents, assign goals, and track work and costs from one place.

It looks like a task manager, but under the hood sit org charts, budgets, governance, and goal alignment. It connects OpenClaw, Claude Code, Codex, Cursor, Gemini CLI, OpenCode, Pi, Hermes, Grok Build, and Kimi Code — the README's rule is that if an agent can receive a heartbeat, it's hired.

Installation

With Node.js 24.11 or newer:

npx paperclipai@latest onboard --yes

The quickstart defaults to trusted local loopback mode. To start in authenticated mode instead, add --bind lan or --bind tailnet. For an isolated trial that never installs a service, use the test-drive path with your provider key:

ANTHROPIC_API_KEY=... npx paperclipai test-drive

You can also run from source: clone the repo, then pnpm install and pnpm dev — the UI and API start at http://localhost:3100 with an embedded PostgreSQL database created automatically. This path needs Node.js 24.11+ and pnpm 9.15+, and — for the native runner build enabled by default — a Rust toolchain (or a prebuilt runner via PAPERCLIP_RUNNER_BINARY).

Real Workflow: Staff an Autonomous Team

1. Define the goal. In the dashboard, state a business outcome — the README's example is "Build the #1 AI note-taking app to $1M MRR."
2. Hire the team. Add agents as CEO, CTO, engineers, designers, or marketers. Each agent gets a role, a reporting line, and its own model and harness.
3. Approve and run. Review the strategy, set company and per-agent budgets, and start the work. Agents wake for assigned tasks, follow-up messages, or configured schedules.

Once the team is running, governance stays in your hands: review and approval stages gate task completion, connection actions can require a one-time approval or a saved rule, and spending is tracked by company, agent, and project with threshold alerts and automatic pauses at configured limits.

Shared skills are managed through Skill Studio: install or write skills for the whole org, test them with saved inputs, and restore earlier versions when an edit goes wrong.

Tips

• Use npx paperclipai test-drive for a throwaway instance already initialized with a CEO agent — each run gets a unique temporary directory unless you pass --data-dir.
• Rerunning onboard keeps your existing config; edit settings later with npx paperclipai configure.
• If npx fails with an E404 because of a private .npmrc registry, force the public one: npx --registry https://registry.npmjs.org paperclipai@latest onboard --yes.
• The team catalog ships preview-ready teams with roles, skills, and routines you can adapt instead of hiring from scratch.
• Usage telemetry is enabled by default; disable it with PAPERCLIP_TELEMETRY_DISABLED=1 or DO_NOT_TRACK=1.

When Not to Use This

For a single prompt or a one-off script, a plain agent is simpler — Paperclip earns its keep when several agents coordinate on ongoing work. It is also not an agent framework: it tells you how to run a company of agents, not how to build one, and it is not a prompt manager — your agents keep their own prompts, models, and runtimes.

See the leaderboard for more skills.