# Add to your Claude Code skills
git clone https://github.com/meetopenbot/openbotLast scanned: 5/30/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-05-30T15:05:01.198Z",
"npmAuditRan": true,
"pipAuditRan": true
}openbot is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by meetopenbot. Meet OpenBot. OS for Agents. It has 310 GitHub stars.
Yes. openbot passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/meetopenbot/openbot" and add it to your Claude Code skills directory (see the Installation section above).
openbot is primarily written in TypeScript. It is open-source under meetopenbot on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh openbot against similar tools.
No comments yet. Be the first to share your thoughts!
OpenBot is a local-first harness for running AI agents. It is built around a small event API, local file storage, and a Melony-powered runtime that routes events to agents and plugins.
~/.openbot.system agent named OpenBot (orchestrator, includes the LLM runtime).state agent for deterministic, non-LLM handling (e.g. /api/state defaults).~/.openbot/agents/<agent-id>/AGENT.md.~/.openbot/plugins.Requires Node.js >=20.12.0.
npm i -g openbot
openbot start
The server listens on http://localhost:4132 by default. Set a different port with:
openbot start --port 3000
For local development:
npm install
npm run dev
OpenBot intentionally keeps the public API small:
GET /api/events opens an SSE stream for a channel or thread.POST /api/publish publishes an event into the harness (defaults to the built-in system agent with the OpenBot / LLM runtime).GET /api/state runs an event and returns the resulting events without opening a stream (defaults to the built-in state agent: storage-oriented plugins, no LLM).You can override the agent with agentId (header, query, or body where applicable).
Example:
curl -X POST http://localhost:4132/api/publish \
-H "content-type: application/json" \
-d '{"type":"agent:invoke","data":{"role":"user","content":"hello"}}'
Useful context can be passed as headers, query params, or body fields:
channelIdthreadIdagentIdrunIdOpenBot reads config from ~/.openbot/config.json.
{
"port": 4132,
"baseDir": "~/.openbot",
"model": "openai/gpt-4o-mini"
}
Variables are read from ~/.openbot/variables.json and applied to the server process environment on startup.
The built-in system agent is always available. Add a custom agent by creating ~/.openbot/agents/<agent-id>/AGENT.md:
---
name: Researcher
description: Helps collect and summarize information.
plugins:
- id: openbot
config:
model: openai/gpt-4o-mini
- id: storage
---
You are a careful research assistant.
Summarize findings clearly and cite sources when available.
Agents are discovered from disk when the server starts.
Built-in plugins include:
storage-toolsdelegationopenbotShared plugins can be placed in ~/.openbot/plugins and referenced by agents.
src/app: CLI, server, event types, and app config.src/harness: orchestration and process helpers.src/plugins: built-in plugin implementations.src/services: local storage service.src/registry: plugin registry.docs: architecture, agents, and plugin notes.Need help or want to share feedback? Join the community on Discord: https://discord.gg/XYYXvN2ebB