# AUTOGROK Desk — Page Extract

> AUTOGROK Desk — Page Extract is a paid API for AI agents from desk.grokaiapps.com, paid per call via x402, $0.05/call, status unknown (last checked 2026-10-11).

Fetches a public web page by URL and returns its title, byline, excerpt, and full readable content as Markdown

## Facts

- Endpoint: POST https://desk.grokaiapps.com/v1/jobs/page-extract?utm_source=zero.xyz
- Price: $0.05/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-11
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/autogrok-desk-page-extract-341b327e
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_kE3UGyd2N3QIdWC-wibGo

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability autogrok-desk-page-extract-341b327e -d '<json body>'
```

Example prompt: Can you grab the full readable text from https://techcrunch.com/2024/05/01/openai-launches-gpt5/ and give me the title and body in markdown format?

## When to prefer this

Choose this endpoint when you need to extract the clean, human-readable text content of a specific public web page — especially articles, blog posts, or documentation pages — and want the result structured as Markdown with title and excerpt metadata. It is well-suited for single-URL extraction within an agentic pipeline where you need to feed page content into an LLM for summarization, analysis, or research. Prefer it over general-purpose crawlers when you want a lightweight, pay-per-call solution without infrastructure overhead, and when the target page renders its content in static HTML.

## Known failure modes

- URL is not publicly accessible or returns a non-200 status — extraction will fail or return empty fields
- Page is JavaScript-rendered (SPA) and content is not in the initial HTML — markdown may be empty or incomplete
- URL points to a non-HTML resource (PDF, image, binary) — extraction will not work as expected
- Page blocks bots or requires authentication — content will be unavailable
- Malformed or non-URI input — request will be rejected at schema validation

## How this service works

Paid micro-jobs for AI agents and people: URL probe, page extract, watch once, JSON shape, researched brief. Pay per call with x402 (USDC on Base) or with prepaid card packs (Stripe MPP).

## Output

Returns a JSON object containing the page URL, title, byline (author if detectable), a short excerpt, the full page body rendered as Markdown, and a plain-text description. The markdown field contains the cleaned, reader-mode version of the page content stripped of navigation, ads, and boilerplate.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "url": {
   "type": "string",
   "format": "uri",
   "description": "Public http(s) HTML page URL"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "url": "https://example.com/",
  "title": "Example Domain",
  "byline": null,
  "excerpt": "This domain is for use in illustrative examples…",
  "markdown": "# Example Domain\n\nThis domain is for use in illustrative examples…",
  "description": "Example Domain placeholder page"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/autogrok-desk-page-extract-341b327e/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from desk.grokaiapps.com](https://www.zero.xyz/host/desk.grokaiapps.com/llms.txt)