the-open-agent/openagent

curl -fsSL https://raw.githubusercontent.com/the-open-agent/openagent/master/scripts/install.sh | bash  (macOS/Linux/WSL)
irm https://raw.githubusercontent.com/the-open-agent/openagent/master/scripts/install.ps1 | iex  (Windows PowerShell, native)
docker-compose up  (Docker)
Build from source: go build (Go 1.25.0+) + cd web && yarn install && yarn start (Node 20+, Yarn 1.x)

Usage

• •Open http://localhost:14000 after the installer starts the server (default port)
• •Connect any model provider and switch per conversation; plug in MCP servers (SSE / Stdio / StreamableHTTP) to expose their tools to agents
• •Use the visual workflow builder, task scheduling, and admin dashboard for usage analytics and tool management
• •Access everything programmatically via the REST API + Swagger UI

Key features

• ✓30+ model providers (OpenAI, Anthropic, Gemini, DeepSeek, Mistral, Grok, Qwen, Ollama, etc.) switchable per conversation
• ✓Autonomous agent loops: browser-use, web search & fetch, shell execution, office (Word/Excel/PPT) automation, transparent tool calls
• ✓RAG knowledge base with document ingestion, semantic search, pluggable embeddings, and isolated stores
• ✓Visual BPMN-style workflow builder with conditional/parallel execution and task scheduling
• ✓Platform features: single binary, native Windows, SSO (OIDC/OAuth2/LDAP/SAML), multi-tenancy, audit logs, admin dashboard

Best for

Teams that want a self-hosted, multi-user AI assistant platform with RAG, agent loops, MCP tool access, and workflow automation behind one dashboard.

Caveats

• ⚠Requires running a server (default port 14000); styled as a full platform rather than a lightweight tool
• ⚠Online demo resets data every 5 minutes (playground); live preview is read-only