agno-agentos

Deploy with AgentOS

Use agno.os.AgentOS to serve agents, teams, and workflows as production APIs. Only requires uv — all dependencies are declared inline via PEP 723.

Quick Start

from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.os import AgentOS

agent = Agent(
    name="My Agent",
    model=Claude(id="claude-sonnet-4-5"),
    instructions="You are a helpful assistant.",
)

agent_os = AgentOS(agents=[agent])
agent_os.serve()

Run with: python app.py — serves at http://localhost:7777.

Full Setup with Agents, Teams, and Workflows

from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.db.sqlite import SqliteDb
from agno.os import AgentOS
from agno.team.team import Team
from agno.tools.calculator import CalculatorTools
from agno.tools.websearch import WebSearchTools
from agno.workflow.step import Step
from agno.workflow.workflow import Workflow

# Database
db = SqliteDb(db_file="tmp/app.db")

# Agents
assistant = Agent(
    name="Assistant",
    model=Claude(id="claude-sonnet-4-5"),
    db=db,
    tools=[CalculatorTools()],
    instructions=["You are a helpful assistant."],
    markdown=True,
    update_memory_on_run=True,
    add_history_to_context=True,
    num_history_runs=3,
)

researcher = Agent(
    name="Researcher",
    model=Claude(id="claude-sonnet-4-5"),
    db=db,
    tools=[WebSearchTools()],
    instructions=["You are a research assistant."],
    markdown=True,
)

# Team
research_team = Team(
    name="Research Team",
    model=Claude(id="claude-sonnet-4-5"),
    members=[assistant, researcher],
    instructions=["Coordinate research tasks."],
    db=db,
    markdown=True,
)

# Workflow
qa_workflow = Workflow(
    name="QA Workflow",
    description="Simple Q&A workflow",
    db=db,
    steps=[Step(name="Answer", agent=assistant)],
)

# AgentOS
agent_os = AgentOS(
    id="my-app",
    description="My production agent application",
    agents=[assistant, researcher],
    teams=[research_team],
    workflows=[qa_workflow],
)

app = agent_os.get_app()  # Returns FastAPI app

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

AgentOS Parameters

AgentOS(
    id="app-id",                          # Unique application ID
    description="What this app does",     # Description
    agents=[agent1, agent2],              # Agents to serve
    teams=[team1],                        # Teams to serve
    workflows=[workflow1],                # Workflows to serve
    knowledge=[knowledge1],              # Knowledge bases (enables upload/search endpoints)
)

Serving Options

# Quick start (blocks, good for development)
agent_os.serve()

# With uvicorn options
agent_os.serve(app="module:app", reload=True)

# Get FastAPI app for custom hosting
app = agent_os.get_app()

# Then run with uvicorn directly:
# uvicorn module:app --reload --port 7777
# Or: fastapi dev module.py

Database Configuration

Development (SQLite):

from agno.db.sqlite import SqliteDb
db = SqliteDb(db_file="tmp/app.db")

Production (PostgreSQL):

from agno.db.postgres import PostgresDb
db = PostgresDb(db_url="postgresql+psycopg://user:pass@host:5432/dbname")

Knowledge Base Registration

Register knowledge bases to enable search and upload endpoints:

from agno.knowledge.knowledge import Knowledge
from agno.knowledge.embedder.openai import OpenAIEmbedder
from agno.vectordb.chroma import ChromaDb

knowledge = Knowledge(
    vector_db=ChromaDb(
        path="tmp/chromadb",
        collection="docs",
        embedder=OpenAIEmbedder(id="text-embedding-3-small"),
    ),
    contents_db=db,  # Required for upload/management endpoints
)

agent = Agent(
    name="Knowledge Agent",
    knowledge=knowledge,
    search_knowledge=True,
)

agent_os = AgentOS(
    agents=[agent],
    knowledge=[knowledge],  # Registers knowledge API endpoints
)

Skills Integration

Load skills into agents served via AgentOS:

from pathlib import Path
from agno.skills import LocalSkills, Skills

skills_dir = Path(__file__).parent / "skills"

agent = Agent(
    name="Skills Agent",
    model=Claude(id="claude-sonnet-4-5"),
    skills=Skills(loaders=[LocalSkills(str(skills_dir))]),
)

agent_os = AgentOS(agents=[agent])

API Endpoints

After starting AgentOS, explore endpoints at:

• http://localhost:7777/docs — Swagger UI
• http://localhost:7777/config — Configuration and available agents/teams/workflows

Anti-Patterns

• Don't skip db= on agents — without storage, sessions don't persist between API calls
• Don't forget contents_db= on Knowledge — required for content upload endpoints
• Don't use SQLite in production — use PostgreSQL for multi-server deployments
• Don't forget extras in PEP 723 header: scripts need agno[os] for AgentOS features
• Don't hardcode secrets — use environment variables for API keys and database URLs

Further Reading

For advanced AgentOS configuration and deployment patterns, read references/api-patterns.md.