Agno

A self-hosted Python AI agent framework and runtime that serves agents through REST and MCP, with database storage and a web control plane.

Agno is a framework and runtime for developers building customer-facing or internal AI agents. It combines agent development, service hosting and a web interface for managing runs on infrastructure you control.

Agents and workflows Build individual agents, cooperating teams and workflows with tools, document knowledge, structured outputs and persistent memory.
Service interfaces AgentOS exposes agents through REST APIs and MCP, and connects them to Slack, Telegram, WhatsApp and Discord.
Managed execution Schedule jobs, run tasks in the background and pause runs for human approval, with durable execution and resumable streaming.
Monitoring and permissions Inspect sessions, traces, evaluations and usage in the control plane, and govern actions with roles, tool permissions and audit logs.

Agno supports your choice of model provider or gateway. AgentOS can also run agents built with LangGraph, DSPy, the Claude SDK or custom code.

The open-source framework uses Apache 2.0. The platform runs locally with Docker, on your servers or in your cloud, and stores sessions, memory, knowledge and traces in your database. Connected external model providers and tools still process requests outside that deployment.

Agno sends a telemetry event for each agent run. You can disable telemetry, and those events don't include prompts, messages or outputs.