What it actually does
Once you have more than one agent doing real work, the problem stops being capability and starts being management. Who is working on what? What did it cost? Who approved it?
Paperclip is a Node.js server with a React UI that orchestrates a team of agents against business goals. You define a goal — the README's example is "build the #1 AI note-taking app to $1M MRR" — then assign roles (CEO, CTO, engineers, designers, marketers) using any bot from any provider, set budgets, review the strategy, and hit go.
It looks like a task manager. Underneath it is org charts, budget enforcement, governance, goal alignment, and coordination between agents. The framing the maintainers use is that you manage business goals, not pull requests.
At 76,000 stars and MIT licensed, it is one of the more serious attempts at the multi-agent management problem rather than another single-agent wrapper.
Who it's for
- 01
Founders running several agents who have lost track of what each one is doing
- 02
Ops leads who need a spend ceiling per workstream, not a surprise invoice
- 03
Small teams trying to get leverage from agents without hiring a manager for them
- 04
Anyone who needs an approval step between an agent's plan and its execution
Where it earns its keep
- Giving a marketing initiative its own agent team with a fixed monthly budget
- Reviewing an agent's proposed strategy before any money or API spend is committed
- Tracking cost per goal so you can tell which automations actually pay for themselves
- Coordinating handoffs between agents instead of running them as disconnected scripts
- Keeping an audit trail of what was decided, by which agent, under whose approval
Use it, or skip it
Reach for it when
- You are running three or more agents and coordination has become the bottleneck
- Budget control matters — you want hard ceilings rather than after-the-fact reporting
- Someone needs to approve plans before agents act on them
- You want one dashboard across providers rather than one console per vendor
Skip it when
- You have one agent doing one job. This is management overhead you do not need yet
- You want agents to run fully autonomously without approval gates — that is not the design
- Your workloads are deterministic pipelines; a scheduler is simpler and more reliable
- You need SOC 2 style assurances from a vendor rather than software you host yourself
10 automations
Ideas, not tutorials. Each one is work a team does by hand today.
- 01Marketing
Budget-capped campaign team
Spin up a marketing agent team with a hard monthly ceiling, so experiments cannot quietly run up an API bill.
- 02Finance
Cost per automation report
Weekly breakdown of spend by goal, so you can retire the automations that cost more than the work they replace.
- 03Operations
Plan approval gate
Require a human sign-off on any agent plan that touches customer-facing systems, with the plan summarised in Slack.
- 04Founders
Quarterly goal decomposition
Feed in the quarter's objective and have the agent team break it into workstreams, then track progress against it.
- 05Marketing
Content pipeline with roles
Researcher, writer, and editor agents in sequence, with the editor's output queued for a human before publishing.
- 06Sales
Sales research desk
A standing agent team that prepares an account brief before every discovery call, budgeted per deal size.
- 07Support
Support triage tier
First-line agents categorise and draft, escalating anything above a confidence threshold to a person.
- 08Engineering
Provider bake-off
Assign the same goal to two agent teams on different providers and compare output quality against cost.
- 09Operations
Runaway detection
Alert when an agent exceeds expected spend or loops without progress, and pause it automatically.
- 10Founders
Weekly org review
Generate a digest of what every agent shipped, what it cost, and what is blocked — the same review you would run with people.
Want one of these running by Friday?
LimeDock builds these as real workflows inside your stack — deployed to your cloud, wired into your Slack and CRM, with the code in your repo. You pay a build fee and your own API keys, nothing else.
Source
Repository stats were read from the GitHub API and reflect the last time we refreshed this entry. The editorial breakdown above is LimeDock’s own analysis — we are not affiliated with paperclipai.
https://github.com/paperclipai/paperclip