Scrape any page.

A small typed action DSL extracts exact fields via html or real Chromium. Interpretation is a separate analyse call. Every run is recorded.

Quickstart Run a scrape

import { Card, CardGrid } from '@astrojs/starlight/components';

<CardGrid stagger>
 <Card title="Quickstart" icon="rocket">
 Sign up, mint an API key, run an html scrape and an analyse call — in under a minute.
 
 [Start →](/docs/quickstart)
 </Card>
 <Card title="Pick an engine" icon="setting">
 HTML for static pages, Browser (real Chromium) for SPAs and clicks. Interpretation is `POST /analyse`, not scrape.
 
 [Compare engines →](/docs/engines)
 </Card>
 <Card title="Actions DSL" icon="document">
 The selector + extractor language: `h1@text`, `many`, nested `output`. CSS is the default — no `css=` prefix.
 
 [Reference →](/docs/actions-dsl)
 </Card>
 <Card title="REST API" icon="seti:json">
 `POST /scrape`, `POST /analyse`, `GET /executions` — full request and response shapes.
 
 [API reference →](/docs/api/scrape)
 </Card>
 <Card title="MCP" icon="puzzle">
 Wire scraper-farm into Claude Code, Cursor, or any MCP-aware agent. Tools: `scrape`, `analyse`, `map`, `docs`, `get_execution`, `list_executions`.
 
 [Set up MCP →](/docs/mcp)
 </Card>
 <Card title="Authentication" icon="seti:lock">
 Bearer tokens, key lifecycle, the `X-Use-Api-Key` override the dashboard uses.
 
 [How auth works →](/docs/authentication)
 </Card>
 <Card title="Pricing & quotas" icon="approve-check">
 Credit cost per engine, monthly plans, concurrency caps, and the 402 / 429 limit responses.
 
 [See pricing →](/docs/pricing)
 </Card>
</CardGrid>

## Smoke test

```bash
curl -sS -X POST https://api.scrapesilo.com/scrape \
  -H "Authorization: Bearer sf_…" \
  -H "Content-Type: application/json" \
  -d '{
        "url": "https://scrapesilo.com/fixtures/article",
        "engine": "html",
        "actions": { "title": "h1", "body": "p", "cta": "a.sf-cta@href" }
      }'
```

Returns an array: `[{ url, data, tookMs, executionId, antibot }]`. The execution row stays around — open `/executions/<id>` in the dashboard to inspect the request, result, and any error.

Markdown agents.txt

Quickstart

Start →

Pick an engine

HTML for static pages, Browser (real Chromium) for SPAs and clicks. Interpretation is POST /analyse, not scrape.

Compare engines →

Actions DSL

The selector + extractor language: h1@text, many, nested output. CSS is the default — no css= prefix.

Reference →

REST API

POST /scrape, POST /analyse, GET /executions — full request and response shapes.

API reference →

MCP

Wire scraper-farm into Claude Code, Cursor, or any MCP-aware agent. Tools: scrape, analyse, map, docs, get_execution, list_executions.

Set up MCP →

Authentication

Bearer tokens, key lifecycle, the X-Use-Api-Key override the dashboard uses.

How auth works →

Pricing & quotas

Credit cost per engine, monthly plans, concurrency caps, and the 402 / 429 limit responses.

See pricing →

Smoke test

curl -sS -X POST https://api.scrapesilo.com/scrape \
  -H "Authorization: Bearer sf_…" \
  -H "Content-Type: application/json" \
  -d '{
        "url": "https://scrapesilo.com/fixtures/article",
        "engine": "html",
        "actions": { "title": "h1", "body": "p", "cta": "a.sf-cta@href" }
      }'

Returns an array: [{ url, data, tookMs, executionId, antibot }]. The execution row stays around — open /executions/<id> in the dashboard to inspect the request, result, and any error.