git-workflow

Build, run, and manage agent platforms.

Introduction

Agno provides software for building, running, and managing agent platforms.

• Build agents using any agent framework.
• Run them as production services with session management, tracing, scheduling, and RBAC.
• Manage your platform using a single control plane.

Agno has a 3-layer architecture. Everything except the control plane is free and open-source.

Example: coding agent as a service

Here's how to run a coding agent as a service.

Built using the Agno SDK

Save this file as workbench.py:

from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.os import AgentOS
from agno.tools.workspace import Workspace

workbench = Agent(
    name="Workbench",
    model="openai:gpt-5.4",
    tools=[Workspace(".",
        allowed=["read", "list", "search"],
        confirm=["write", "edit", "delete", "shell"],
    )],
    enable_agentic_memory=True,
    add_history_to_context=True,
    num_history_runs=3,
)

# Serve using AgentOS → streaming, auth, session isolation, API endpoints
agent_os = AgentOS(
    agents=[workbench],
    tracing=True,
    db=SqliteDb(db_file="agno.db")
)
app = agent_os.get_app()

Workspace(".") scopes the agent to the current directory. read, list, and search run freely; write, edit, delete, and shell require human approval.

from agno.agents.claude import ClaudeAgent
from agno.db.sqlite import SqliteDb
from agno.os import AgentOS

agent = ClaudeAgent(
    name="Claude Agent",
    model="claude-opus-4-7",
    allowed_tools=["Read", "Bash"],
    permission_mode="acceptEdits",
)

agent_os = AgentOS(agents=[agent], db=SqliteDb(db_file="agno.db"), tracing=True)
app = agent_os.get_app()

Run it

uv pip install -U 'agno[os]' openai

export OPENAI_API_KEY=sk-***

fastapi dev workbench.py

In 30 lines of code, you get:

• A FastAPI backend with 50+ endpoints
• Streaming responses, persistent sessions, per-user isolation
• Cron scheduling, human approval flows, and RBAC
• Native OpenTelemetry tracing

API is available at http://localhost:8000 and OpenAPI spec at http://localhost:8000/docs.

Manage your platform using the AgentOS UI

You can use the AgentOS UI to manage your agent platform. Use it to test your agents, inspect runs, view traces, manage sessions, and monitor the health of the system. It's free to use with a local AgentOS.

1. Open os.agno.com and sign in.
2. Click "Connect OS".
3. Select "Local" to connect to a local AgentOS.
4. Enter your endpoint URL (default: http://localhost:8000).
5. Name it "Local AgentOS" and click "Connect".

Open Chat, select your agent, and ask:

Tell me more about the project and the key files

The agent reads your workspace and answers grounded in what it actually finds. Try a follow-up like "create a NOTES.md with three key takeaways." The run pauses for your approval before the file is written, since write is in the confirm list.

https://github.com/user-attachments/assets/adb38f55-1d9d-463e-8ca9-966bb6bdc37a

Get started

Choose whichever path suits you best:

• Read the docs
• Build your first agent
• Start from a template
  • Coda → A code companion that lives in Slack and works alongside your team.
  • Dash → A self-learning data agent that grounds answers in your business context.
  • Scout → An agent that navigates information sources like Slack, Google Drive, and Notion to assemble answers.
• Start from a blank canvas. Build on top of the leanest agent platform template.

Agno features

• Production API. 50+ endpoints with SSE and websockets make it easy to build a product on top of your agent platform.
• Storage. Store sessions, memory, knowledge, and traces in your own database. Use postgres for quick read/write data like sessions and memory. Use clickhouse for OLAP data like traces.
• 100+ integrations. Integrate with 100+ tools using pre-built toolkits.
• Context Providers. Use strategies like context providers to access live data stored in Slack, Drive, wikis, MCP, and custom sources.
• Human approval. Built in mechanisms for pausing runs for user confirmation up to blocking tools that require admin approval.
• Observability. Get monitoring via OpenTelemetry tracing, run history, and audit logs out of the box.
• Security. Get JWT-based RBAC and multi-user, multi-tenant isolation out of the box.
• Interfaces. Expose your agents via Slack, Telegram, WhatsApp, Discord, AG-UI, A2A.
• Scheduling. Cron-based scheduling and background jobs with no external infrastructure.
• Deploy anywhere. Run on any cloud platform that can run a containerized image. Docker, Railway, AWS, GCP.

Coding agent integration

Two options for using Agno with your coding tools:

1. Add Agno documentation as an indexed source.

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

   Also works with VSCode, Windsurf, and similar tools.

2. Add Agno documentation as an MCP server.

   Add docs.agno.com/mcp as an MCP server to your favourite coding agent.

Contributing

See the contributing guide.

Telemetry

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

↑ Back to top