OpenAgent
OpenAgent is a self-hostable, single-binary personal AI assistant powered by LLMs, RAG, and agent loops. It supports MCP tools, 30+ model providers, Telegram, Discord, and WeCom, with no vendor lock-in.
Visit the-open-agent/openagentOverview
OpenAgent is an open-source, self-hostable personal AI assistant that combines LLMs, a RAG knowledge base, and autonomous agent loops in a single-binary platform. It supports MCP tools, 30+ model providers, and multi-channel messaging including Telegram, Discord, and WeCom.
Key Features
- Single-binary deployment: download and run natively on Windows, macOS, or Linux without an installer, Docker, or WSL.
- Full MCP support lets agents use tools, call APIs, and reason across multi-step workflows with persistent memory.
- Unified access to 30+ model providers, including OpenAI, Claude, Gemini, DeepSeek, Qwen, Mistral, and Ollama.
- RAG knowledge base ingests documents and retrieves context via vector search with pluggable embeddings and isolated stores.
- Multi-channel gateway deploys one agent across Telegram, Discord, and WeCom simultaneously.
- Admin dashboard provides usage statistics, activity monitoring, tool management, and request logs.
Use Cases
- Build a self-hosted personal assistant that can browse the web, run shell commands, and call MCP-compatible tools.
- Create a RAG knowledge base from PDFs, Word documents, and spreadsheets for semantic search and context-aware answers.
- Deploy a single agent across Telegram, Discord, and WeCom for multi-platform chat access.
- Automate multi-step workflows with a visual BPMN-style builder, conditional and parallel execution, and scheduling.
Getting Started
- Run the provided installer for macOS, Linux, or WSL, or the PowerShell installer for Windows.
- Open the local dashboard on port 14000 after installation.
- Alternatively, use Docker Compose or build from source with Go 1.25.0+ and Node.js 20+ with Yarn 1.x.
Deployment & Requirements
- Pre-built binaries are available for Linux, macOS, and Windows on x86_64 and arm64.
- Windows runs natively without WSL or Docker.
- Docker deployment is available via docker-compose up.
- Building from source requires Go 1.25.0+ for the backend and Node.js 20+ with Yarn 1.x for the frontend.
Before You Adopt
- License: Apache-2.0. Review its terms before using, modifying, or distributing the project.
- Self-hosting keeps data on the user's machine, which is relevant for privacy-sensitive deployments.
- The public playground resets data every five minutes, so it is not intended for persistent work.