Agno

Agno is a framework and runtime for building agent platforms. You write agents with the Agno SDK, serve them with the AgentOS runtime, and manage the platform through the AgentOS web UI. The design goal is ownership of the agent stack: your data, memory, and security posture stay in your own infrastructure.

A starter template sets up a REST API for serving agents, a Postgres database for data and traces, an MCP server, and a control plane, all running locally in Docker. The templates are identical apart from their deploy scripts and target Railway, Docker, AWS, GCP, Azure, Fly, Render, Modal, or Helm. The documentation is also exposed as an MCP server and as an llms-full.txt index for use inside coding agents.

Features

• Production API: 50+ endpoints with SSE and websockets for building a product on top of your agents
• Storage: sessions, memory, knowledge, and traces persist in your own database
• Toolkits: 100+ pre-built integrations, including GitHub, Slack, and Postgres
• Context providers: live data from Slack, Drive, wikis, MCP, and custom sources
• Human approval: runs pause for user confirmation, and tools can be blocked pending admin approval
• Observability: OpenTelemetry tracing, run history, and audit logs
• Security: JWT-based role-based access control with multi-user, multi-tenant isolation
• Interfaces: agents exposed through Slack, Telegram, WhatsApp, Discord, AG-UI, and A2A
• Scheduling: cron-based schedules and background jobs without external infrastructure
• Telemetry control: one event per agent run, with no prompts or outputs sent, disabled by AGNO_TELEMETRY=false