agno 2.4.4

Agno: a lightweight library for building Multi-Agent Systems

Build, run, manage multi-agent systems.

Docs  •  Cookbook  •  Community  •  Discord

What is Agno?

Agno is a framework, runtime, and control plane for multi-agent systems.

Layer	What it does
Framework	Build agents, teams, and workflows with memory, knowledge, guardrails, and 100+ integrations
AgentOS Runtime	Run your system in production with a stateless, secure FastAPI backend
Control Plane	Test, monitor, and manage your system using the AgentOS UI

Why Agno?

• Private by design. AgentOS runs in your cloud. The control plane connects directly to your runtime from your browser. No retention costs, no vendor lock-in, no compliance headaches.
• Production-ready on day one. Pre-built FastAPI runtime with SSE endpoints, ready to deploy.
• Fast. 529× faster instantiation than LangGraph. 24× lower memory. See benchmarks →

Example

An agent with MCP tools, persistent state, served via FastAPI:

from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.models.anthropic import Claude
from agno.os import AgentOS
from agno.tools.mcp import MCPTools

agno_agent = Agent(
    name="Agno Agent",
    model=Claude(id="claude-sonnet-4-5"),
    db=SqliteDb(db_file="agno.db"),
    tools=[MCPTools(transport="streamable-http", url="https://docs.agno.com/mcp")],
    add_history_to_context=True,
    markdown=True,
)

agent_os = AgentOS(agents=[agno_agent])
app = agent_os.get_app()

if __name__ == "__main__":
    agent_os.serve(app="agno_agent:app", reload=True)

Run this and connect to the AgentOS UI:

https://github.com/user-attachments/assets/feb23db8-15cc-4e88-be7c-01a21a03ebf6

Features

Core

• Model-agnostic: OpenAI, Anthropic, Google, local models
• Type-safe I/O with input_schema and output_schema
• Async-first, built for long-running tasks
• Natively multimodal (text, images, audio, video, files)

Memory & Knowledge

• Persistent storage for session history and state
• User memory across sessions
• Agentic RAG with 20+ vector stores, hybrid search, reranking
• Culture: shared long-term memory across agents

Orchestration

• Human-in-the-loop (confirmations, approvals, overrides)
• Guardrails for validation and security
• Pre/post hooks for the agent lifecycle
• First-class MCP and A2A support
• 100+ built-in toolkits

Production

• Ready-to-use FastAPI runtime
• Integrated control plane UI
• Evals for accuracy, performance, latency
• Durable execution for resumable workflows
• RBAC and per-agent permissions

Getting Started

1. Follow the quickstart guide
2. Browse the cookbook for real-world examples
3. Read the docs to go deeper

Performance

Agent workloads spawn hundreds of instances. Stateless, horizontal scalability isn't optional.

Metric	Agno	LangGraph	PydanticAI	CrewAI
Instantiation	3μs	1,587μs (529×)	170μs (57×)	210μs (70×)
Memory	6.6 KiB	161 KiB (24×)	29 KiB (4×)	66 KiB (10×)

_{Apple M4 MacBook Pro, Oct 2025. Run benchmarks yourself →}

https://github.com/user-attachments/assets/54b98576-1859-4880-9f2d-15e1a426719d

IDE Integration

Add our docs to your AI-enabled editor:

Cursor: Settings → Indexing & Docs → Add https://docs.agno.com/llms-full.txt

Also works with VSCode, Windsurf, and similar tools.

Contributing

We welcome contributions. See the contributing guide.

Telemetry

Agno logs which model providers are used to prioritize updates. Disable with AGNO_TELEMETRY=false.

↑ Back to top