Extract clean Markdown from a URL

Fetch a single URL and return its main content as clean Markdown, with the navigation, ads, cookie banners and boilerplate stripped. This is the synchronous endpoint: one URL in, extracted content out, in a single round trip. For many URLs use `POST /bulk`; to follow links from a starting page use `POST /crawl`. `format` selects the output shape — `markdown` (default), `json` (typed blocks), `chunks` (pre-split for embedding), `html`, or `links`. `json` and `chunks` are Pro+ features, as is `render_js`, which runs the page through a headless browser first for client-side-rendered sites. Costs one request against your monthly quota.

Authorization

BearerAuth

AuthorizationBearer <token>

Your API key, sent as Authorization: Bearer wm_.... Keys are wm_ followed by 40 hex characters. Create one with POST /register (free, no auth) or POST /keys.

In: header

Request Body

application/json

TypeScript Definitions

Use the request body type in TypeScript.

allow_domains?array<string>|

Restrict this request to these domains. Narrowing only — listing a domain your key doesn't already permit is silently ignored, never a widening. Null leaves the key's own policy in force.

deny_patterns?array<string>|

Additional URL patterns to refuse for this request, on top of the key's own deny list. Narrowing only.

format?Format

Output shape. markdown (default) returns prose; json returns typed blocks; chunks returns pre-split passages ready for embedding; html returns cleaned HTML; links returns just the outbound links. Exactly one response field is populated per format. json and chunks are Pro+.

Default"markdown"

Value in

"markdown"
"json"
"chunks"
"html"
"links"

render_js?Render Js

Run the page through a headless browser before extracting, for sites that render their content client-side. Slower. Pro+ only — a Free key gets 403 plan_not_supported.

Defaultfalse

respect_robots?|

Override robots.txt handling for this request. lax→strict is honoured; a strict→lax downgrade is silently ignored, since that would widen the key's policy.

retry?Retry

Extra attempts if the fetch times out. 0 (default) means a single attempt. Each retry opens a fresh connection. Timeouts only — a 4xx from the target or an empty page is deterministic and is never retried. On this synchronous endpoint each timed-out attempt costs 20-30s, so large values belong on /bulk.

Range0 <= value

Default0

url*Url

The page to extract. Must be http(s) and publicly reachable.

Formaturi

Length1 <= length <= 2083

Response Body

`application/json`

curl -X POST "https://example.com/extract" \  -H "Content-Type: application/json" \  -d '{    "url": "http://example.com"  }'

{  "blocks": [    {      "level": 0,      "text": "string",      "type": "heading"    }  ],  "chunks": [    {      "end_token": 0,      "start_token": 0,      "text": "string"    }  ],  "html": "string",  "links": [    "string"  ],  "markdown": "string",  "metadata": {    "author": "string",    "date": "string",    "retrieved_at": "2019-08-24T14:15:22Z",    "title": "string",    "url": "string"  },  "metrics": {    "content_bytes": 0,    "input_tokens": 0,    "output_tokens": 0,    "reduction_pct": 0,    "tokens_saved": 0  },  "request_id": "266ea41d-adf5-480b-af50-15b940c2b846"}

Extract clean Markdown from a URL

Authorization

Request Body

Response Body

200application/json

401application/json

402application/json

403application/json

422application/json

429application/json

`application/json`

`application/json`

`application/json`

`application/json`

`application/json`

`application/json`