Paperclip via Claude Code: multi-agent orchestration locally

My notes

Summary

Paperclip is a multi-agent orchestration framework that runs locally and auto-creates a team of AI agents (starting with a CEO agent) that each trigger their own onboarding tasks. Unlike single-agent tools like OpenClaw, Paperclip coordinates multiple agents that understand the codebase and business context, essentially automating manual delegation workflows.

Key Insight

• Paperclip positions itself as the “next level up” from single-agent tools like OpenClaw: instead of one agent doing tasks, you get an orchestrated team of agents with distinct roles
• The setup involves forking the Paperclip repo and running it locally, connecting it to existing GitHub repositories so agents can review and understand current projects
• Each agent, once created, automatically triggers a familiarisation task on itself, a self-onboarding pattern where agents bootstrap their own context
• The CEO agent starts first and then builds out the rest of the team, mimicking a top-down org structure
• The key value proposition is compressing time: tasks that would take a founder hours or days of manual delegation and execution happen autonomously in parallel
• Scheduling capability means this can run as a persistent background process, not just on-demand
• Early stage and experimental: the creator is still in the experimenting phase, so stability and real output quality are unknown