What problem it solves

AI agents need controlled access to browsers, files, and tools while people retain visibility and control over their actions. OpenBot addresses this with isolated Bot computers, policy evaluation, audit records, and human takeover for sensitive steps.

Key features

• Isolated computers for every Bot — Each Bot can receive a dedicated container with its own Chromium instance, browser profile, logins, and workspace. Bots can also run shell commands in that workspace through the same governed action path.

  Source: README, agent-computer

• Policy checks and an audit trail — The gateway evaluates actions before they reach a Bot computer and records the decision first. CEL rules can inspect tool, actor, page, command, file, MCP, and initiator details, while audits identify permitted, refused, and failed actions.

  Source: README, server

• Bring any AG-UI agent — OpenBot accepts endpoints that speak AG-UI, allowing existing agents to become Bots without being tied to a particular framework. Agent endpoints receive target validation, and authorization headers can be stored without being returned.

  Source: README, examples

• Human oversight and takeover — People can watch a Bot's screen and activity, answer requests for help, and take control when manual interaction is required. The handoff and release events are recorded, and Bot actions are refused while a person is driving.

  Source: README, app

• Self-hosted deployment and durable context — The project can run from source with Docker Compose or as a published container, with PostgreSQL storing workspace configuration. Conversations and deployment-owned threads remain available through CopilotKit Intelligence.

  Source: README, docker-compose.yml

Quick start

Copied verbatim from the project's README.

   cp .env.example .env
   bun install

   npx --yes copilotkit@latest login
   npx --yes copilotkit@latest project select
   bun scripts/setup-learning.ts

   bash scripts/start.sh

Who it may suit

Inference: OpenBot may suit developers and teams that want to run AG-UI agents locally, add governed browser and file workflows, and retain operational control over agent actions.

Topics to discuss

• Which governance model best fits a workplace where people can take over while Bots remain independently accountable?

• How should policy rules balance useful automation with predictable refusals and understandable audit records?

• Which AG-UI agent integrations and component workflows should the community prioritize next?

Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.

Read the original README https://github.com/CopilotKit/OpenBot#readme

ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.