> ## Documentation Index
> Fetch the complete documentation index at: https://docs.opengeni.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# For AI agents

> The opengeni-client skill, machine-readable docs, and what to treat as authoritative when integrating OpenGeni.

If you are a coding agent integrating OpenGeni into a product, start with the `opengeni-client` skill. It covers the default embed, onboarding, tools, isolation, background work, verification, and the decisions to ask your user about.

## Install the skill

One command, from the root of the product repository:

```bash theme={null}
npx skills add Cloudgeni-ai/opengeni --skill opengeni-client
```

`bunx skills add Cloudgeni-ai/opengeni --skill opengeni-client` works the same way. Without that CLI, copy the directory with git:

```bash theme={null}
git clone --depth 1 --filter=blob:none --sparse https://github.com/Cloudgeni-ai/opengeni /tmp/opengeni
git -C /tmp/opengeni sparse-checkout set .agents/skills/opengeni-client
mkdir -p .agents/skills && cp -R /tmp/opengeni/.agents/skills/opengeni-client .agents/skills/
```

Without git, read the whole skill as plain Markdown at [`https://docs.opengeni.ai/reference/opengeni-client-skill.md`](/reference/opengeni-client-skill). It mirrors every file of the skill, and a CI check keeps it identical to the repository copy.

The skill is for the agent building the integration. Do not attach it to the customer-facing OpenGeni agent.

## Machine-readable docs

| Resource | Contents |
| - | - |
| [`/llms.txt`](https://docs.opengeni.ai/llms.txt) | Index of every page with a one-line description |
| [`/llms-full.txt`](https://docs.opengeni.ai/llms-full.txt) | Every page in one file |
| Any page URL + `.md` | That page as Markdown, for example `/embed-manually.md` |
| `https://docs.opengeni.ai/mcp` | An MCP server that searches these docs |

## What is authoritative

When sources disagree, trust them in this order:

1. **The live deployment.** `GET /v1/config/client` and `GET /v1/access/me` show what the target deployment supports and what the credential may do.
2. **The installed packages.** The exported types of the installed `@opengeni/sdk` and `@opengeni/react` define the exact API you can call.
3. **The skill and the canonical repository docs**, such as [`docs/product-integration.md`](https://github.com/Cloudgeni-ai/opengeni/blob/main/docs/product-integration.md).
4. **These pages**, which summarize the above for people.

## Where to start reading

* [How it fits together](/integrate/how-it-fits-together) for the architecture.
* [Embed manually](/embed-manually) for the default integration code.
* [SDK](/reference/sdk) and [React components](/reference/react) for the API surface.