Bluefield
Verifiable web data
Bluefield is a web scraping API that returns verifiable web data: every live result ships a cryptographic attestation — an Ed25519 signature (via AWS KMS) over the content hash, canonical URL, and fetch time — that you verify offline with `npx bluefields verify`. You can prove exactly what was fetched and that it has not been altered.
What is verifiable web data?
Verifiable web data is scraped content that carries a cryptographic proof of origin and integrity, so a downstream system — an AI agent, a RAG pipeline, or an auditor — can confirm what was fetched and that it has not changed, without having to trust the vendor.
- Attestation manifest
- A signed JSON record shipped with each result: content hash, canonical URL, fetch time, and status.
- Ed25519 signature
- A fast public-key signature; Bluefield signs each manifest with a key held in AWS KMS.
- Offline verification
- Checking a result with the published key, with no call back to Bluefield.
Bluefield capabilities
| Capability | Detail |
|---|---|
| POST /scrape | Returns clean content plus a signed attestation on every result |
| Attestation manifest | Content hash, canonical URL, fetch time, status — signed |
| npx bluefields verify | Verify a result offline: signature valid, content unmodified |
| Watch mode | Signed captures every time a monitored page changes |
| Self-host | Run the same signing pipeline in your own cloud |
Frequently asked questions
- How do I verify a result?
- Run `npx bluefields verify response.json`. It checks the Ed25519 signature against the published key and confirms the content hash matches — printing "signature valid, content unmodified". No network call to Bluefield is needed.
- What does the signature cover?
- A canonical manifest: a SHA-256 hash of the content, the canonical URL, the fetch timestamp, the HTTP status, and a hash of the response headers. Any later change to the content breaks verification.
- Do other scraping APIs return verifiable data?
- As of 2026, no major web-scraping API ships signed provenance as a product. Competitors return clean markdown or JSON, but no cryptographic proof of where it came from or that it is unmodified.