Skip to main content
SSSuperScraper
Shipping notes

Changelog

What shipped, newest first. Updated as features land.

  1. Improved

    Status page wired to a real health check

    The status page now runs a live probe against the production API on every load and reports Responding or Unknown from the result. The dashboard row says Not monitored, because no probe covers it yet. The graphite dark theme from the design spec is now the shipped default.

  2. New

    API deployed to production

    The API runs on Railway and answers GET /v1/status with 200, reporting per-subsystem health for Supabase, Redis, and the job queues. Database migrations 001 through 057 are applied.

  3. Improved

    Site rebuilt on the evidence-terminal design system

    The hero proof block renders a real captured response instead of an illustration, and every remaining unverifiable claim was rewritten or removed.

  4. New

    /benchmarks page

    The anti-bot benchmark (96.3%, 26/27, measured 2026-06-23) with full methodology, per-vendor results, the exact reproduce command, and an explicit what-this-does-not-prove section.

  5. New

    Docs rebuilt around the agent path

    New IA: Quickstart, Authentication (incl. POST /v1/keys/provision), Core endpoints, Agents (MCP, SKILL.md, llms.txt, framework adapters), Reference (errors, rate limits, webhooks), and a Firecrawl migration page with an explicit gap table.

  6. New

    Hosted MCP server

    Point Claude, Cursor, Codex, Cline, or any MCP-aware tool at mcp.superscraper.dev/{key}/mcp for live web access with one URL of config. Zero install, Streamable HTTP. The hosted endpoint goes live at launch.

  7. New

    Agent self-provisioning

    POST /v1/keys/provision lets an agent mint its own free-tier key, and /agent-onboarding/SKILL.md teaches it the API. No human in the loop.

  8. New

    /v1/parse for documents

    A PDF or DOCX, by upload or URL, becomes markdown a model can read. SSRF-guarded.

  9. Improved

    robots.txt respected by default

    Every fetch honors robots.txt out of the box, with an explicit per-request opt out for content you are authorized to access.

  10. New

    Provenance on structured results

    Structured results now ship with a confidence score, the extraction method, and the fetch tier that won. A black box becomes a decision you can audit.

  11. New

    Multi-endpoint playground

    Run scrape, extract, map, and more from the browser with no key and no card, and see the confidence score and source trail on the output.

Try the latest in the playground

Everything here is live. Paste a URL and see the output with a confidence score and a source trail.

Open the playgroundGet an API key