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

# LLM Ingestion

> How to make OrgX docs consumable by LLMs with Mintlify and code-generated references.

## Built-in Mintlify LLM Surfaces

Mintlify already exposes LLM-friendly formats:

| Surface                     | What it provides                                         |
| --------------------------- | -------------------------------------------------------- |
| `/llms.txt`                 | Compact index of available docs pages for model crawlers |
| `/llms-full.txt`            | Expanded combined content export                         |
| `<page>.md`                 | Markdown rendering for any docs page URL                 |
| `/prompt.txt` / `/skill.md` | Optional prompt/skill context files when configured      |

That means you do **not** need a custom parser just to support copy-to-LLM workflows.

## Grounding Docs in Code

The reliability requirement is drift control: docs should be generated from source, not hand-maintained.

For OrgX cloud MCP tools:

```bash theme={"dark"}
pnpm docs:generate
```

For OpenClaw plugin MCP tools (cross-repo generation):

```bash theme={"dark"}
pnpm docs:generate-openclaw-plugin
```

This command reads:

* `Code/orgx-openclaw-plugin/src/tools/core-tools.ts`
* `Code/orgx-openclaw-plugin/src/mcp-http-handler.ts`

and regenerates:

* `docs/mintlify/generated/openclaw-plugin-tool-catalog.json`
* `docs/mintlify/api/openclaw-plugin-mcp-tools.mdx`

## Recommended LLM-Ready Workflow

1. Regenerate catalogs before each docs release.
2. Commit generated outputs with timestamps.
3. Run broken-link checks:
   ```bash theme={"dark"}
   cd docs/mintlify
   npx mintlify@latest broken-links
   ```
4. Spot-check `.md` variants for key pages (for example `/api/mcp-tools.md`).
5. Verify `llms.txt` and `llms-full.txt` include newly added pages after deploy.

## Copy-to-LLM for Every Page

For end users, the simplest "copy to LLM" path is:

* open the page in docs
* append `.md` to the URL
* copy the markdown output

Example:

* Docs page: `/api/openclaw-plugin-mcp-tools`
* LLM-ready variant: `/api/openclaw-plugin-mcp-tools.md`