MarkUDown

Async jobs — MarkUDown web scraping API docs

Async Jobs

Long-running operations are processed asynchronously. The API returns an id immediately; poll the status endpoint until the job completes.

Submit an async job

POST /api/crawl
X-API-KEY: your-api-key

{
  "url": ["https://docs.example.com/"],
  "max_depth": 3,
  "limit": 100
}

// Response
{
  "success": true,
  "id": "abc123",
  "url": "/api/crawl/abc123"
}

Poll for results

GET /api/crawl/abc123
X-API-KEY: your-api-key

// Waiting / running ("progress" appears on some jobs):
{ "success": false, "id": "abc123", "status": "pending" }
{ "success": false, "id": "abc123", "status": "processing", "progress": 50 }

// When complete:
{
  "success": true,
  "id": "abc123",
  "status": "completed",
  "data": {
    "success": true,
    "total": 42,
    "data": [...]
  }
}

// On failure:
{ "success": false, "id": "abc123", "status": "failed", "error": "job failed: ..." }

// Result no longer kept (see below):
{ "success": false, "id": "abc123", "status": "not_found" }
// A job id that is not yours: HTTP 404 { "detail": "Job not found" }
Supported async endpoints:
crawlmapextractprompt-extractsearchrssbatch-scrapechange-detectiondeep-researchagentsmart-extractrankdataset

Prompt Extract jobs are polled at /api/extract/{id}. Finished results are kept only for a limited time (at most 1–2 hours, less under heavy load) — read them promptly or set callback_url; after that the status is not_found. Only the API key that submitted a job can read it.

Cancel a job

Crawl, Extract, Agent, Smart Extract and Dataset jobs can be cancelled with DELETE on the same URL.

DELETE /api/crawl/abc123
X-API-KEY: your-api-key

{ "success": true, "id": "abc123", "status": "cancelled" }
// or, when the job can no longer be removed:
{ "success": false, "id": "abc123", "status": "not_cancellable" }

Webhook (callback_url)

Endpoints that accept callback_url POST this JSON to it when the job ends. The URL must be a public http(s) address.

{
  "event": "completed",          // or "failed"
  "queue": "crawl",
  "jobId": "abc123",
  "data": { ... },               // same as "data" when polling; "error" instead on failure
  "timestamp": "2026-11-16T14:30:00.000Z"
}