OpenAgent

Next-generation personal AI assistant powered by LLM, RAG and agent loops,
supporting computer-use, browser-use and coding agent

OpenAgent is an open-source personal AI assistant that brings together powerful LLMs, your own knowledge base, and autonomous agent loops — all in one self-hostable platform. Connect any model provider, build a RAG knowledge base from your documents, and let agents browse the web, run code, and call any MCP-compatible tool on your behalf.

Online Demo

Quick Start

Pre-built binaries are available for Linux, macOS, and Windows (amd64 / arm64). The install script downloads the latest release, installs it, and starts the server on port 14000.

Install binary (recommended)

macOS / Linux / WSL

curl -fsSL --proto '=https' --tlsv1.2 \
  https://raw.githubusercontent.com/the-open-agent/openagent/master/scripts/install.sh | bash

Windows (PowerShell)

irm https://raw.githubusercontent.com/the-open-agent/openagent/master/scripts/install.ps1 | iex

Then open http://localhost:14000.

Optional environment variables: OPENAGENT_VERSION, INSTALL_DIR, BIN_DIR.

Build from source

# Backend
go build

# Frontend
cd web && yarn install && yarn start

Highlights

Agent Loops

• Browser-Use — drive a real browser: navigate, click, fill forms, scrape, and screenshot pages
• Web Search & Fetch — search the web and pull page content directly into the agent's context
• Shell Execution — run shell commands and scripts from within the agent loop
• Office Automation — read and write Word, Excel, and PowerPoint files
• MCP (Model Context Protocol) — connect any MCP-compatible server over SSE, Stdio, or StreamableHTTP and expose its tools to the agent
• Transparent Tool Calls — see exactly which tool was invoked, with what arguments, and what it returned, step by step

RAG & Knowledge Base

• Document Ingestion — upload PDFs, Word docs, Excel sheets, and more; they are chunked, embedded, and indexed automatically
• Semantic Search — every chat retrieves the most relevant passages from your knowledge base before the LLM responds
• Pluggable Embedding Providers — OpenAI, Azure, Gemini, Qwen, Cohere, Jina, HuggingFace, local models, and more
• Per-Store Isolation — organise knowledge into separate stores and assign them to individual chats or applications

30+ Model Providers

Works out of the box with all major LLM providers — configure as many as you like and switch between them per chat:

OpenAI · Azure OpenAI · Claude (Anthropic) · Gemini (Google) · DeepSeek · Mistral · Grok · Qwen · Doubao · Moonshot · ChatGLM · Baichuan · Ernie · iFlytek · HuggingFace · Cohere · Amazon Bedrock · OpenRouter · local models · and more

Workflow Automation

• Visual Workflow Builder — compose multi-step pipelines with a BPMN-style editor
• Conditional & Parallel Execution — branch on gateway conditions and run tasks concurrently
• Task Scheduling — run workflows or agent jobs on a recurring schedule
• Usage Analytics — track token consumption and cost per provider, model, and user

Platform Features

• Single Sign-On — OIDC / OAuth2 / LDAP / SAML via the integrated auth layer
• Multi-tenant — separate workspaces per user or organisation
• REST API + Swagger UI — every feature is accessible programmatically
• Audit Logs — full activity history for every action
• File & Video Management — built-in storage for files, images, and video content

Documentation

https://www.openagentai.org/

Community

• Discord: https://discord.gg/5rPsrAzK7S
• Issues and PRs are welcome — please open an issue first to discuss larger changes

License

Apache-2.0