Drop a CSV or JSON file, click the value you used, and name the tool input it feeds. The page builds an ocg-document-span@1 sidecar record per mapping: the SHA-256 of the file bytes, the exact cell or RFC 6901 pointer, the extracted string, and the declared normalization rule. Export the batch as a spans file and attach it at verify time on the Artifact Verifier's tie-out block. Everything runs in this page; the file never leaves your browser.
Four checks run per span, in order: the SHA-256 of the supplied bytes equals the recorded document digest; the named locator resolves in a parse of the same bytes by the pinned RFC 4180 parser (or an RFC 6901 walk for JSON); the extracted string equals the recorded lexical byte for byte; and the canonical digest of the normalized lexical equals the canonical digest of the value at the input pointer, when that value was supplied. The state marks:
The extractor is pinned: , checked against the published kernel digest list vendored as-of . Records naming any other digest are refused. Export the batch and attach it at verify time on the Artifact Verifier's tie-out block, which accepts the same records next to a receipt. Primary text: RFC 4180 (CSV) and RFC 6901 (pointers).