Guide · Agent skill
How to use the Delegate Work to Paperclip Agents skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The delegate-work skill teaches an AI assistant to pass a piece of work to an agent that already exists in a Paperclip team. It is aimed at team leads, developers and operators who run several AI agents in Paperclip and want to assign work from a chat. After this guide you will know what the skill does, how to install it, how to phrase a request that triggers it, and what to check before any work starts.
Delegate Work to Paperclip Agentspaperclipai/paperclip · MIT Download the skill Quick start
- 1Install the plugin and connect it. Install the Paperclip plugin and connect it to your assistant as paperclip_connection. Confirm the connection identifies your person and company.
- 2Confirm your team and agents exist. Make sure the team has agents configured in Paperclip, with model credentials and an execution budget set up there.
- 3Download the skill ZIP. Download the delegate-work ZIP from this page.
- 4Install the skill. Unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 5Make a clear delegation request. Describe the objective, the context you want included, any constraints and the expected deliverable.
- 6Resolve questions and confirm. Answer the assistant's questions about assignee, team or scope, and accept the note that delegation can use the team's execution budget.
- 7Track the returned task. Save the task ID and link, then ask the assistant to check progress later through paperclip_read_task.
What the delegate-work skill makes the assistant do
Agent skills are instruction files that give an AI assistant a repeatable procedure for one kind of task. This skill covers delegation: taking a request from the user and passing the work to an existing Paperclip agent with a durable task reference.
Before creating anything, the assistant identifies the connected person and company, lists the agents and, where relevant, the projects. It picks the agent suited to the outcome and asks the user to resolve any unclear team, scope or assignee. If an agent is paused or unavailable, the assistant reports it and does not promise the agent will run.
- Checks the connected person and company through the paperclip_connection plugin
- Lists agents and projects, then selects the most suitable agent
- Prepares a task with objective, user-provided context, constraints and expected deliverable
- Creates the task once, with a UUID requestId so retries are safe
- Returns the task ID and link, then reads progress later with paperclip_read_task
Install the delegate-work skill
Download the ZIP for the skill from the page you are reading. The skill is a single file, SKILL.md, inside the delegate-work folder of the paperclip repository.
For a personal install, unzip the download into ~/.claude/skills/. To limit the skill to one project, unzip it into that project's .claude/skills/ folder instead.
You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Set up Paperclip before you delegate
The skill only hands work to agents that already exist. It does not create teams, and it does not supply model credentials. Installing the Paperclip plugin and connecting it to the assistant (paperclip_connection) is required. Setup and approval decisions are handled in Paperclip's own UI.
Model credentials and the execution budget are configured separately in Paperclip. If they are missing, delegation will fail or produce nothing useful.
- A Paperclip plugin connected to the assistant as paperclip_connection
- An existing Paperclip team with agents configured
- Model credentials and execution budget set up in Paperclip
Trigger the skill with natural requests
You do not need a special command. Ask the assistant in plain language to hand work to a Paperclip agent. The more specific the objective, context and expected output, the less the assistant has to ask you back.
When the scope is clear, your request to delegate counts as authorization to start the work. Expect the assistant to warn you that delegation can start paid autonomous agent work using the team's execution budget.
- "Delegate a competitor pricing summary to our research agent. Deliverable: a one-page table with sources."
- "Have the Paperclip agent for the blog fix the broken links in the March draft and report back."
- "Assign the onboarding checklist review to whichever agent handles operations in our main team."
Check status and track the task
After creating a task, the assistant returns the durable task ID and a link. It distinguishes submitted, running, blocked and completed states based on the evidence Paperclip returns, not on guesses.
The task continues after the chat closes. Ask the assistant to check progress and it will use paperclip_read_task. The skill is designed not to poll continuously, so expect a status check when you ask for one.
- Keep the task ID to find the work again in Paperclip
- If a call fails with an unknown outcome, the assistant inspects tasks before retrying rather than creating a duplicate
Safety notes: scripts and network access
This skill runs no scripts. It uses the network because it talks to Paperclip through the connected plugin to list agents, create tasks and read progress.
Because each delegated task can start autonomous agent work and consume the team's budget, review the objective and constraints before confirming. Use the assistant to check which agent is selected when the team has several.
- Runs scripts: no
- Uses network: yes, to reach Paperclip through the connected plugin
- Retries reuse the same requestId and identical arguments, so a retry does not create a second task
What people use it for
- Asking a research agent to compile a sourced summary while you continue other work
- Sending a bug-fix or content-revision task to the agent that owns that area of the team
- Assigning a recurring review to an operations agent with a clear deliverable and a traceable task ID
- Recovering safely from a dropped connection by checking existing tasks instead of resubmitting
Frequently asked questions
Is the delegate-work skill free?
The source does not state a price or licence terms for the skill beyond its listing. The item is listed under the MIT licence. Delegated tasks can start paid autonomous agent work using your team's execution budget, so check your Paperclip budget before delegating.
How do I install the delegate-work skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
What do I need before I can use it?
You need a Paperclip plugin connected to the assistant as paperclip_connection, an existing Paperclip team with agents configured, and model credentials and an execution budget set up separately in Paperclip. The skill does not create teams or provide credentials.
Why does delegation fail or do nothing?
The most likely causes are no configured Paperclip team, missing model credentials or budget, or a paused or unavailable agent. The assistant is meant to report paused or unavailable agents rather than promise they will run. Check setup in Paperclip's UI.
Does the skill run scripts or access the internet?
It runs no scripts. It uses the network to communicate with Paperclip through the connected plugin.
Should I use this instead of asking the assistant directly?
Use it when you want work done by an agent in your Paperclip team, with a tracked task. For simple one-off questions the assistant can answer itself, delegation adds cost and is not needed.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits team leads and developers who already run AI agents in Paperclip and want assignments made from a chat with a traceable task record. It is not a good fit if you have no configured Paperclip team, or if you only need quick answers, since each delegation can start paid agent work from your execution budget.
Download the skill