SuperAGI is an open-source framework for developers who want to build, manage, and run autonomous AI agents. Agents run concurrently, are extended with toolkits that connect them to external systems, store memory between runs, and are operated through a graphical interface with an action console for giving input and granting permissions.
You can run it locally with Docker Compose (including a GPU compose file for local LLMs), deploy it to DigitalOcean, or use the hosted SuperAGI Cloud by signing in with GitHub and adding a model-provider API key. A marketplace supplies toolkits, and a Postman collection documents the API.
Features
- Agent lifecycle: provision, spawn, and deploy autonomous agents
- Toolkits: marketplace toolkits including Twitter, Instagram, GitHub, Jira, Notion, email, Google Calendar, Google Search, DuckDuckGo, SerpAPI, Searx, DALL-E, a web scraper, a file manager, a coding tool, and Apollo
- Graphical interface: manage agents from a web UI
- Action console: interact with running agents by supplying input and permissions
- Vector databases: connect agents to multiple vector stores
- Performance telemetry: insight into agent runs for tuning
- Token controls: limit token usage to manage cost
- Agent memory: stored memory lets agents adapt across runs
- Models: custom fine-tuned models for business-specific use cases
- Workflows: predefined ReAct steps for task automation
