Choosing an endpoint
Pick the right endpoint when two look alike — scrape or fetch, which bulk API, schema or scrape, which monitor type, and search or sitemap.
For AI agents: https://bytekit.com/llms.txt — every docs page is available as markdown by appending .md
Several ByteKit endpoints overlap. Each choice below says which one to use and links to the guide with the detail.
Scrape or fetch
Use /v1/scrape when you want a JSON envelope with the content as raw, markdown, links or images, plus metadata, or when you need rendering options or async.
Use /v1/fetch when you want the target's response as the body of a successful reply instead of a JSON envelope, with metadata in X-Fetch-* headers; it takes no format, and a scrape format such as markdown returns 400 validation_error that points at /v1/scrape (details).
Which bulk API
Use /v1/bulk when the job includes screenshots, alone or together with scrapes.
Use /v1/scrape/bulk for a job of scrapes and /v1/fetch/bulk for a job of raw fetches.
Schema or scrape
Use /v1/schema when you want only data from a page as a JSON object that follows a JSON Schema you supply; today a successful extraction costs 10 credits plus the page's bandwidth (Cost).
Use /v1/scrape when you want the page's content itself, or when you also want that object in the same response, by adding a json entry to formats; the entry costs 10 credits, charged only when the extraction validates, on top of the scrape's own bandwidth charge (json entry).
Screenshot monitor or scrape monitor
Set type to screenshot, the default, to detect visual change: it compares perceptual hashes of successive captures and alerts when the change reaches change_threshold (Monitors).
Set type to scrape to detect content change: it alerts when the content hash differs from the previous capture's (Monitors).
Search or sitemap
Use /v1/search to find URLs from a query; it returns ranked results and costs credits.
Use /v1/sitemap to discover the URLs on one domain; it starts a job whose results you download as a file.