openagent

About

Personal AI assistants usually demand Python environments, Docker, or per-seat cloud accounts, which puts them out of reach for privacy-conscious individuals and small teams. OpenAgent compiles a Go backend with a React frontend into a single binary that installs via curl or PowerShell script, runs natively on Linux, macOS, and Windows without WSL or Docker, and starts a web UI on port 14000. Its agent loop draws on 30-plus switchable LLM providers (OpenAI, Anthropic, Gemini, DeepSeek, Mistral, Qwen, OpenRouter, Ollama, and more), executes tools through any MCP-compatible server over SSE, stdio, or Streamable HTTP, and adds browser automation, shell execution, office automation, and a RAG knowledge base with pluggable embeddings. Multi-tenancy with OIDC/OAuth2/LDAP/SAML, audit logs, and usage-cost analytics target self-hosted team deployment. It suits users who want a private, single-binary assistant combining coding-agent abilities with office and browser automation.