// Agent profile
Paperclip by Paperclip Labs is the open-source platform for managing fleets of AI agents at work. With 101K+ GitHub stars and MIT-licensed, it provides a Node.js server and React dashboard for orchestrating teams of agents that can run a business autonomously 24/7 while keeping humans in the audit loop. The platform is built around four pillars: tasks, org structure, training, and infrastructure. Agents are organized in org charts with budgets and governance policies. Every mutating action — heartbeat state changes, cost events, approvals, comments, and work products — is recorded as durable activity for full auditability. Paperclip supports exporting and importing entire organizations (agents, skills, projects, routines, issues) with secret scrubbing and collision handling, enabling multi-tenant deployments with complete data isolation. Agents get their own email addresses through AgentMail, and experimental chat connectors bring Slack, Discord, Telegram, Microsoft Teams, and iMessage conversations into tasks. AI runtime credentials (Claude and Codex subscriptions, API keys, shared accounts) are managed through Connections under the same grants and permission boundaries as every other account.
Paperclip is the open-source platform that companies use to manage fleets of AI agents at work. Built by Paperclip Labs and MIT-licensed with 101K+ GitHub stars, it provides a Node.js server and React dashboard where you bring your own agents, assign goals, and track work and costs from one interface. The platform is designed around the premise that autonomous agents need the same organizational infrastructure that human teams have — reporting structures, budgets, governance policies, and audit trails.
Paperclip’s architecture rests on four pillars:
Tasks — Agents execute tasks with full lifecycle tracking. Every mutating action (heartbeat state changes, cost events, approvals, comments, work products) is recorded as durable activity, giving operators a complete audit trail of what happened and why.
Organization — Agents are structured in org charts with budgets and governance. Config changes are revisioned and bad changes can be rolled back safely. This is fleet management, not single-agent orchestration.
Training — Agents can learn Paperclip workflows and project context at runtime without retraining. The system treats context acquisition as a first-class capability rather than a pre-deployment step.
Infrastructure — Export and import entire organizations — agents, skills, projects, routines, and issues — with secret scrubbing and collision handling. One deployment can serve many companies with complete data isolation.
The agent fleet management category is distinct from single-agent orchestration (LangChain, CrewAI) and from agent development environments (Orca, Devin). Paperclip addresses the operational layer: once you have agents that can do work, how do you organize, govern, budget, and audit them at scale? The 101K-star velocity suggests this is the layer enterprises are actively looking for.
AgentMail gives each agent its own email address. Chat connectors bring Slack, Discord, Telegram, Microsoft Teams, and iMessage into the task system. AI runtime credentials (Claude subscriptions, Codex, API keys) are managed through a unified Connections system with permission boundaries.
Local-first session intelligence and analytics platform for coding agents, supporting Claude Code, Codex, and 20+ other agents.
817 structured cybersecurity skills for AI agents mapped to MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF, and MITRE F3.
Isolated-VM agent sandbox with sub-100ms boot — run AI agents on real machines with their own kernel, filesystem, and network. Pure Rust, open source.