bluefield
Changelog

What we have shipped

Product updates in reverse chronological order.

  1. New

    Per-call attestation on one-shot fetches

    • POST /scrape now returns a signed, offline-verifiable attestation manifest on every live fetch — not just stored watch-mode captures.
    • Verify any result offline with the published spec and the bluefields verify CLI; the production key is pinned in a trust store.
    • Cached responses and PDFs are unsigned in this first version.
  2. New

    Credit billing, MCP server, and web search

    • All async API calls now consume credits per a published tariff. One page scrape costs 1 credit.
    • Screenshots and PDF parsing each add 1 credit.
    • The MCP server is available. Register Bluefield with any MCP-compatible agent host to give your agents scrape, map, crawl, and search as native tools.
    • POST /search performs a web search and optionally scrapes result pages in one call.
    • Four self-serve plans are available on the pricing page.
  3. New

    Structured extraction and batch scrape

    • POST /extract accepts a URL and a JSON schema and uses an LLM to return structured data matching your schema.
    • POST /batch/scrape accepts up to 1,000 URLs and processes them as a single background job.
    • SSE streaming is available on Growth and Scale plans via GET /crawl/:id for live progress.
  4. New

    Watch mode and webhook delivery

    • Subscribe to any URL with POST /subscriptions. Bluefield polls on an adaptive schedule and fires a signed webhook when content changes meaningfully.
    • The adaptive scheduler adjusts poll frequency based on observed change cadence for each page.
    • Webhook secrets rotate with a 24 hour overlap window so deploys do not drop events.
    • GET /history returns the full change log for any subscription.
  5. New

    Scrape API, crawl, and map

    • POST /scrape returns clean markdown, HTML, links, screenshots, or structured JSON from any URL.
    • Headless rendering handles JavaScript-heavy pages. Supports waitFor, named sessions, cookies, and page actions.
    • POST /crawl recursively scrapes a site up to configurable depth and page limits.
    • POST /map discovers all accessible URLs on a site in one call.
    • Watch mode stored captures include a signed attestation manifest for independent verification.