Read any web page
web_fetch is a $0.20 GET request on the Wuthering AI API. Fetch a page as clean markdown or raw HTML, including sites that block bots. It answers complete JSON, and a call that fails is not charged.
Endpoint
GET /v1/web_fetch
Price per successful call
$0.20
Failed calls
Not charged
curl 'https://wutheringai.com/v1/web_fetch?url=…&token=YOUR_TOKEN'
What does web_fetch return, and when should you call it?
Fetch a URL and return its content, as readable markdown by default or as raw HTML when you need the markup.
Handles pages that block automated access, which is what makes competitor pricing and documentation pages reachable at all. Choose markdown for reading and analysis; choose html only when structure matters, such as extracting a pricing table's cells or finding hidden elements, because html costs far more context.
Returns: the page content as text in the requested format.
Use to read a specific known page. Don't use to find pages, which is wuthering_web_search.
What parameters does web_fetch take?
web_fetch accepts 2 parameters, of which 1 is required. Array values are comma-separated on the query string.
| Parameter | Type | Required | Description |
|---|---|---|---|
| url | string | yes | The full URL to fetch, including protocol. If you do not have the exact URL, find it with your own web search first (for example: site:acme.com pricing), or with wuthering_web_search if you have none. A guessed URL returns someone's 404 page as though it were content. |
| format | string | no | markdown returns readable text and is the right default. html returns raw markup and costs substantially more context; use it only when you need the structure.One of: markdown, htmlDefaults to markdown |
Which playbooks call web_fetch?
5 of the research playbooks name web_fetch in their method. Each is free to read and says where in the sequence the call belongs.
- Where is the unoccupied position in this category?Map the category around one domain as an idea maze — the forks it has already run, who took each branch, which walls are still walls and which have moved — then return the one open position with the offer that takes it: buyer, bundle, price metric and message, decided together.
- Who are we really up against?Work April Dunford's positioning sequence from public evidence: the alternatives a prospect would actually use, what only you do, the value that follows, who cares most, and the category that makes all of it obvious.
- What do we charge, and does the money work?Anchor price against what competitors actually charge and meter on, size willingness to pay from what the problem costs today, then run CAC, LTV and payback across three scenarios.
- What should we build next?Gather the language of people switching into and out of this category, code every quote to the force that drives it, size each theme against real search demand, and return a 50/50 roadmap.
- What would prove us wrong?Turn a belief the plan depends on into an explanation that cannot be bent to fit any result, name the rival explanations that would produce the same evidence, and pay only for the observations that would come out differently under each.
Reference for web_fetch · last updated · published by Wuthering AI