# Agno — Python SDK, self-hosted AgentOS runtime and hosted control plane for building and operating agent platforms.

> Source: The Agents Index — https://theagentsindex.com/agno (structured, researched, re-verified)
> Facts last verified: 2026-10-09

Agno is an agent platform made of three parts: a Python SDK for agents, teams and workflows, the open-source AgentOS runtime you deploy in your own cloud, and a browser control plane that connects to it. Agents are served over REST, MCP and chat interfaces.

| Fact | Value |
| --- | --- |
| Website | https://www.agno.com |
| Price | from $150/month |
| Free tier | Yes |

Free is a standing $0/month plan: unlimited control plane features against AgentOS instances on your own machine, unlimited usage and retention, serving over REST, MCP and chat at any scale. It excludes live AgentOS connections, team seats, SAML SSO and audit logs. Pro adds a 14-day free trial on top, started from the control plane.

## What it does

**Writes code** · Yes — Ships a CodingTools toolkit whose documented job is editing files and running commands in a scoped project directory.

> Build coding agents that inspect repositories, edit files, run commands, and verify changes.
> — https://docs.agno.com/use-cases/coding-agents · Page subtitle under the Coding Agents heading

**Debugs and fixes** · Yes — Bug fixing is a documented pattern: reproduce, patch, rerun the failing tests.

> Reproduce a failure, make a focused change, and rerun the failing tests.
> — https://docs.agno.com/use-cases/coding-agents · Common patterns table, Bug fixing row

**Writes tests** · Partly — The documented Verify stage runs the repository's existing gates; test authoring appears only in the example prompt, not as a capability claim.

> Run the repository's tests, lint, type checks, and build inside the execution environment.
> — https://docs.agno.com/use-cases/coding-agents · Define the engineering loop table, Verify row

**Reviews code** · Yes — Code review over a diff is one of the documented coding-agent patterns.

> Inspect a diff, run targeted checks, and return prioritized findings.
> — https://docs.agno.com/use-cases/coding-agents · Common patterns table, Code review row

**Whole-codebase work** · Partly — Repository-wide migrations are a named pattern, but the docs expect unresolved cases to be reported rather than fully landed.

> Apply repeatable changes across files and report unresolved cases.
> — https://docs.agno.com/use-cases/coding-agents · Common patterns table, Migrations row

**Ships it** · Partly — Publishing is deliberately outside the default loop; commits, PRs and deploys must be exposed as separately gated tools.

> Keep commits, pushes, pull requests, and deployments outside the default loop or expose them as separately gated tools.
> — https://docs.agno.com/use-cases/coding-agents · Define the engineering loop table, Publish row

**Done-for-you service** · Partly — You run the software yourself, but Enterprise adds vendor engineers on the build: dedicated help with your product agent, monthly office hours and ChatGPT/Claude app help.

> DEDICATED HELP BUILDING YOUR PRODUCT AGENT
> — https://www.agno.com/pricing · Enterprise plan card, ADD-ONS list

## How much it does unattended

**Runs autonomously** · Yes — Agents, teams and workflows run unattended on cron schedules and as background jobs, with persisted retries, so a run needs no human present unless a pause is configured.

> AgentOS stores schedules and run history in the platform database, invokes existing agent, team, or workflow endpoints, and lets agents manage schedules through `SchedulerTools`.
> — https://docs.agno.com/features/scheduling · Scheduling, opening paragraph

**Multi-agent** · Yes — Teams are a first-class primitive with a delegating leader, nestable sub-teams and four coordination modes, so several agents can work one task together.

> A Team coordinates agents or nested teams. In the default coordinate mode, the leader can delegate tasks based on member roles and synthesize their results.
> — https://docs.agno.com/teams/overview · What are Teams?, opening paragraph

**Agent permissions** · Yes — Permissions are enforced at runtime per agent and per tool, and a tool can be flagged to pause for user confirmation or for an admin approval that leaves a record.

> You can flag a tool to require user confirmation before it executes, to pause for user input or to have execution handled outside Agno.
> — https://www.agno.com/products/sdk · FAQ, 'Does Agno support human in the loop and admin approval?'

## Which models it runs on

**Claude** · Yes — Anthropic is a registered provider key in the SDK's model resolver, with separate adapters for Claude on Bedrock, Vertex AI and Azure AI Foundry.

> `anthropic`, `cohere`, `dashscope`, `deepseek`, `google`, `google-interactions`, `inception`, `internlm`, `llama-openai`, `meta`, `minimax`, `mistral`, `moonshot`, `open-responses`, `openai`
> — https://docs.agno.com/models/providers/model-index · Model String Key Catalog table, Provider APIs row

**GPT** · Yes — OpenAI is a registered provider key with both Chat Completions and Responses adapters, plus a separate Azure OpenAI deployment adapter.

> `anthropic`, `cohere`, `dashscope`, `deepseek`, `google`, `google-interactions`, `inception`, `internlm`, `llama-openai`, `meta`, `minimax`, `mistral`, `moonshot`, `open-responses`, `openai`
> — https://docs.agno.com/models/providers/model-index · Model String Key Catalog table, Provider APIs row

**Open models** · Yes — Open-weight models run through documented local runtime adapters for Ollama, vLLM, llama.cpp and LM Studio, so inference can stay on your own hardware.

> `llama-cpp`, `llmman`, `lmstudio`, `ollama`, `ollama-responses`, `vllm`
> — https://docs.agno.com/models/providers/model-index · Model String Key Catalog table, Local runtimes row

**Model choice** · Yes — You choose the provider and the model, passed as a string or a class, and fallback models plus custom model classes are supported; the vendor supplies no model of its own.

> Native support for 30+ model providers. Agno handles the API quirks for you.
> — https://www.agno.com/ · 'Your agent platform belongs to you' section, 'Use any model, provider or gateway' card

## Where you use it

**In your editor** · Partly — There is no Agno IDE extension; instead the CLI registers your running AgentOS as an MCP server inside Cursor, Claude Code and Codex, which is a connection rather than an editor surface.

> CONNECT YOUR AGENTOS TO CLAUDE CODE, CODEX AND CURSOR
> — https://www.agno.com/pricing · Free plan card, feature list

**On the command line** · Yes — A first-party terminal tool ships with the SDK and also standalone as agnoctl: it scaffolds projects, runs them, connects MCP clients and manages service-account tokens.

> The `agno` command scaffolds AgentOS projects, runs them with Docker Compose, connects your coding agents over MCP, and manages service-account tokens.
> — https://docs.agno.com/cli/overview · Agno CLI overview, paragraph under the first code block

**In your pipeline** · No — The enumerated serving surfaces are REST, MCP in Claude/ChatGPT and chat apps; no pull-request or CI surface appears.

> Serve your agents in your product, AI apps (ChatGPT, Claude) or chat apps (Slack, Discord). Bring your agents everywhere using REST, MCP or Interfaces.
> — https://www.agno.com/ · How it works, step three 'Serve your agents anywhere'

**In a browser** · Yes — The hosted control plane is a browser app for chat, sessions, traces, evals and approvals, and its Studio canvas builds and publishes agents without code.

> Studio is a canvas built into the Control Plane: drag, drop and publish an agent, and it gets a live API endpoint.
> — https://www.agno.com/products/control-plane · FAQ, 'Can I build AI agents in the Agno Control Plane without writing code?'

## Whose machine it runs on

**Self-hosted** · Yes — The runtime and database are yours to host anywhere a container runs; only the control plane UI is hosted by default, and a self-hosted control plane is an Enterprise item.

> Run in any cloud, on-premise or locally Your agent platform runs anywhere you can run a container. You get full control.
> — https://www.agno.com/ · Homepage, 'Your agent platform belongs to you' section, fourth card

**Open source** · Partly — The SDK and AgentOS runtime are Apache-2.0 on GitHub, while the control plane is a hosted product whose self-hosted variant is quoted per deal on Enterprise.

> Agno is distributed under the [Apache-2.0 license](LICENSE).
> — https://github.com/agno-agi/agno · Repository README, License section

## What it costs to run

**How it meters** · Yes — Billing is per team seat and per live AgentOS connection, flat; nothing inside a connection is metered, and model tokens and compute are billed by your own providers.

> There are no per-event fees, no token fees and no egress fees. You pay for seats and live connections.
> — https://www.agno.com/pricing · FAQ, 'Does Agno charge per event, per token or for data egress?'

**Free tier** · Yes — A standing $0 plan builds and runs the platform at any scale and gives the full control plane against a local AgentOS, with no sales call required to start.

> Free $0/month
> — https://www.agno.com/pricing · Free plan card, price cell

**API access** · Yes — Every agent you register is exposed over a REST API with 80+ endpoints and an OpenAPI spec, served from your own deployment rather than a vendor-hosted endpoint.

> AgentOS turns your agents into a production-grade FastAPI application: 80+ endpoints, MCP server, request isolation, JWT-based RBAC, durability, tracing.
> — https://www.agno.com/products/runtime · 'A production API and MCP server' card

**MCP server** · Yes — One flag serves an MCP endpoint beside the REST API, with eight built-in tools for discovering and running components; ChatGPT and Claude connect as clients on any plan.

> Set `mcp=True` to serve an MCP endpoint at `/mcp` alongside the AgentOS REST API.
> — https://docs.agno.com/features/mcp-server · AgentOS MCP Server, opening sentence

**Bring your own key** · Yes — You supply provider credentials per model integration and run the whole stack on your own infrastructure, so inference keys and compute are both yours.

> Install the provider integration and set its credentials before running the example:
> — https://docs.agno.com/models/overview · What are Models?, paragraph above the install snippet

## Buying it for a team

**A company can buy it** · Yes — Pro is sold to a team with three seats and roles included, Enterprise is quoted for organisations, and there is a dedicated enterprise contact route.

> Pro starts with three team seats and one live connection for $150 per month. Everyone on those seats shares access to the same connected runtimes.
> — https://www.agno.com/pricing · FAQ, 'How many members and live connections does Agno Pro include?'

**Seat model** · Yes — Pro bundles three seats into the $150 and charges $30 per month for each extra one; no ceiling is published, and live connections are a second, separately priced unit.

> Yes. Extra seats are $30 per month each. Extra live connections are $95 per month each.
> — https://www.agno.com/pricing · FAQ, 'Can I add more seats or connections to Agno Pro?'

**Pooled budget** · Partly — There is no metered balance to pool; seats share access to the same connected runtimes and usage inside a connection is unlimited.

> Everyone on those seats shares access to the same connected runtimes.
> — https://www.agno.com/pricing · FAQ, 'How many members and live connections does Agno Pro include?'

**Admin controls** · Yes — Roles, per-agent and per-tool scopes, guardrails and admin approvals are enforced by the runtime; Pro gives basic RBAC and per-resource scoping is Enterprise.

> Role-based access control, per-agent and per-tool permissions, user and admin approvals, audit logs and guardrails. Enforced at runtime.
> — https://www.agno.com/ · 'Governance built for agents' section body text

**Audit log** · Yes — Administrative and agent actions are logged to your own database rather than the vendor's; end-user audit logs are the Enterprise-tier extension of this.

> Every action logged to your database: who did what, when, and to which agent.
> — https://www.agno.com/products/control-plane · 'Governance enforced at runtime' section, Audit trail card

**Single sign-on** · Partly — SAML SSO is a $300 per month add-on on Pro and included on Enterprise; nothing on the site names OIDC as a supported protocol.

> Pro includes the Owner, Admin, Member and Viewer roles, and SAML SSO is available on Pro as an add-on. SAML SSO is included on Enterprise, along with custom RBAC with per-resource scoping.
> — https://www.agno.com/pricing · Frequently asked questions, 'Does Agno support SSO and role-based access control?'

## What happens to your code

**Opt out of training** · Yes — There is nothing to opt out of by design: the browser talks straight to your runtime, so the vendor states it never holds a copy of your data at all.

> The control plane reads from your runtime directly in the browser. Agno never holds a copy of your data.
> — https://www.agno.com/pricing · FAQ, 'Where does Agno's AgentOS store my data, and how long do you keep it?'

**Data residency** · Yes — Jurisdiction is whatever you choose, because the runtime is a container in your own VPC using your own database; the vendor holds no copy to locate.

> In your own database, inside your own infrastructure. AgentOS runs as a container in your VPC.
> — https://www.agno.com/pricing · FAQ, 'Where does Agno's AgentOS store my data, and how long do you keep it?'

**Getting out** · Yes — Exit is trivial because state never leaves your database, retention is yours to set, and no credit balance exists that could expire on cancellation.

> You keep everything until you delete it, so retention is infinite and entirely your call.
> — https://www.agno.com/pricing · Frequently asked questions, 'Where does Agno's AgentOS store my data, and how long do you keep it?'

**Certifications** · Partly — SOC 2 compliance is asserted in the trust band on the enterprise and SDK pages, but no ISO 27001 or GDPR statement is published; security reviews are an Enterprise item.

> SOC 2 COMPLIANT
> — https://www.agno.com/enterprise · Trust statistics band below the contact form

## Platform record (as of 2026-08-30)

Reported by the code-hosting platform, not by the vendor.

| Fact | Value |
| --- | --- |
| Licence | Apache-2.0 |
| Stars | 41969 |
| Forks | 5844 |
| Last push | 2026-08-30 |

## Provenance

1 of 27 sources behind the facts above are independent of the vendor.

## Coverage elsewhere

None recorded. Every fact above comes from the vendor and is not independently corroborated yet.

## What changed

- 2026-10-09 — Capabilities · Commercial terms · Free tier
- 2026-10-07 — Capabilities · Commercial terms · Free tier
- 2026-09-02 — Capabilities · Commercial terms · Free tier
- 2026-08-31 — Capabilities · Commercial terms · Free tier · Price
- 2026-08-23 — Alternatives · Documentation link · One-liner · Summary
- 2026-08-19 — Summary

## Correcting this record

Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Agno, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth.

Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.