# Machine-readable documentation (https://www.scrape-sdk.com/docs/agents/machine-readable-docs)

Canonical: https://www.scrape-sdk.com/docs/agents/machine-readable-docs
Markdown: https://www.scrape-sdk.com/docs/agents/machine-readable-docs.md



The docs site exposes the active navigation through formats that do not require scraping rendered HTML.

| Endpoint            | Purpose                                |
| :------------------ | :------------------------------------- |
| `/llms.txt`         | Site-level LLM index                   |
| `/docs/llms.txt`    | Documentation-only index               |
| `/llms-full.txt`    | Combined current documentation text    |
| `/feeds/docs.jsonl` | One typed JSON object per current page |
| `/schemamap.xml`    | Machine-readable surface map           |
| `/docs/<path>.md`   | Raw Markdown for one current page      |

Every rendered page includes **Copy Markdown** and **View Markdown** controls.

## Prefer canonical URLs [#prefer-canonical-urls]

Use current `/docs/...` URLs. Append `.md` to get the page as Markdown, for example `/docs/quickstart.md`.

## Retrieve one page [#retrieve-one-page]

```bash
curl https://www.scrape-sdk.com/docs/quickstart.md
```

The response starts with the page title and canonical URL, then the Markdown body.

## Retrieve the full corpus [#retrieve-the-full-corpus]

Use `/llms-full.txt` for a compact single request or `/feeds/docs.jsonl` when your indexer wants discrete route records.

```bash
curl https://www.scrape-sdk.com/llms.txt
curl https://www.scrape-sdk.com/feeds/docs.jsonl
```

## Coding-agent skill [#coding-agent-skill]

```bash
npx skills add SoulSniper-V2/scrape-sdk --skill scrape-sdk
```

Agents should refresh `/llms.txt` (or this page's Markdown) before integrating the package, instead of scraping the HTML docs.