CopilotKit/OpenBot

Open-source AI coworkers that each get a computer of their own: a browser, files and tools, with every action decided before it happens and recorded after. Bring any AG-UI agent.

What it solves

OpenBot provides a self-hosted agent platform that allows companies to deploy and manage AI coworkers without relying on third-party hosted services. It solves the problem of trust and control by ensuring that the entire agent stack—including the browser, files, and data—runs on the company's own infrastructure, with strict governance and audit trails for every action the AI takes.

How it works

OpenBot acts as a template for a private AI assistant environment. It uses the AG-UI protocol to connect to agents built on any framework (like LangGraph or CrewAI). Each agent is given its own isolated container with a dedicated browser profile and workspace. A central gateway intercepts every tool call, file access, or browser action, evaluating it against a CEL-based policy to either permit or deny the action before it is recorded in an audit log.

Who it’s for

It is designed for companies and engineering teams who want to deploy AI agents with full ownership of their infrastructure, strict security boundaries, and the ability to customize the agent's capabilities and governance rules.

Highlights

  • Isolated Environments: Each bot gets its own Docker container, browser profile, and shell access.
  • Strict Governance: A fail-closed gateway ensures all actions are checked against security policies and recorded in an audit trail.
  • Human-in-the-Loop: Users can "take the wheel" to handle 2FA or login walls, then hand control back to the bot.
  • Framework Agnostic: Supports any agent via the AG-UI protocol, regardless of the underlying framework.
  • Rich Interactions: Bots can respond with React components instead of just text.
  • Governed MCP: Integrated support for Model Context Protocol (MCP) servers for tools like Google Drive and Notion.

Related

  • Project
    Anil-matcha/open-dots

    An open-source, self-hosted personal AI agent workspace that provides a governed action gateway for secure tool use and an optional computer runtime.

  • Project
    YaoApp/yao

    A self-hosted, cross-device platform for managing AI agents and workspaces, featuring a task board and an open API for workflow integration.

  • Project
    nightly-labs/openbot

    A local-first desktop workspace for persistent AI teammates, consolidating multiple agent providers like Claude Code and Codex into a single interface with dedicated workspaces.

  • Project
    obot-platform/obot

    Obot is an open‑source governance platform that sits between AI‑enabled tools and LLM or tool‑calling services. It provides centralized credential storage, per‑user access policies, sandboxed hosting of custom tool servers, device‑side auditing (Obob Sentry), and correlated audit logs, enabling enterprises to securely manage and monitor AI usage across desktops and cloud services.

  • Project
    HKUDS/OpenOPC

    An AI-native company framework that automates the recruitment, orchestration, and learning of role-specific AI agents to execute complex end-to-end projects.