Coordinate agent teams and governed work in Paperclip
Use Paperclip as a self-hosted control plane for assigning goals, tasks, budgets, approvals, routines, and run history across multiple AI agents.
npx skills add agentskillexchange/skills --skill coordinate-agent-teams-and-governed-work-in-paperclip
Use Paperclip when an operator needs to manage a working group of AI agents as an auditable team rather than a pile of separate chat or terminal sessions. The workflow is concrete: deploy Paperclip, create an organization, register human users and agent identities, connect supported runtimes such as OpenClaw, Claude Code, Codex, Cursor, Gemini CLI, OpenCode, HTTP, process, or MCP-backed adapters, then assign goals, tasks, routines, budgets, permissions, approvals, and artifacts from the Paperclip dashboard.
What this skill actually does
Invoke this instead of using each agent product normally when the work requires persistent task threads, shared organizational context, cost controls, approval gates, scheduled routines, run history, and cross-agent delegation that a single agent UI cannot provide. It is especially useful when a team has many simultaneous coding or operations agents and needs one place to see who owns what, what ran, what it cost, and what still needs review.
This is not a generic product card or broad orchestration-framework listing. The scope is the repeatable operator workflow for running Paperclip as an agent-team control plane: onboarding with the documented CLI, creating the first company and agent, connecting approved adapters, and using Paperclip’s task, heartbeat, governance, skills, routine, artifact, and budget surfaces to supervise agent work. Keep model selection, downstream coding, and external service actions bounded by the connected agent/runtime permissions and Paperclip approval rules.
Inputs and prerequisites: Node.js 24 or later, pnpm/corepack as needed by upstream setup, Paperclip CLI package, a local machine or server/VPS deployment target, and whichever agent runtimes or adapter credentials the operator chooses to connect.
Setup notes: install and start locally with npx paperclipai onboard –yes, then complete onboarding at the local Paperclip instance. For team/server use, follow the upstream Server / VPS installation path and configure HTTPS, storage, database, identity, adapter, and secret settings according to the docs.
Source and verification boundary: use https://docs.paperclip.ing/guides/getting-started/installation/ and https://github.com/paperclipai/paperclip as the canonical references before running the workflow; keep deployment commands, adapter configuration, permissions, and generated outputs reviewable against upstream documentation.
Framework fit: publish this as Multi-Framework because upstream is intentionally agent-runtime agnostic and documents support for multiple agent harnesses and adapter types. Do not map it narrowly to MCP unless the specific use case is only exposing Paperclip-connected operations through an MCP adapter.