Every Sourced figure on an intelligence-grade asset traces to a named primary document — an issuer filing, technical report, or company release with a date. This archive maps figure → document → retrieval → preservation, so a reader (or a court) can check the substrate, not take our word for it. Once byte-capture is complete it becomes the link-rot protection and the defamation shield for the Sourced tier; at v0.1 it is the provenance map that byte-preservation is built on. Read the machine feed: /api/v1/archive-manifest.json.
v0.1 — spine, honestly scoped.provenance captured for 100% of intelligence-grade Sourced figures (31 figures across the 6 assets), zero silent gaps. byte-hash pending — cryptographic SHA-256 of each source file's bytes is captured on desktop from the preserved document; large technical-report PDFs cannot be byte-preserved in the build container. Until byte-capture is complete for all documents, this page does not claim "every figure links to a preserved, hashed copy" — no claim ships that the archive cannot yet back. The Kamoa-Kakula 2025 MRE primary release was content-verified against its figures on 2026-07-05.
What is recorded per figure
primary_document issuer · title · type · date · canonical URL
retrieval content-verified / url-recorded / unfetched — with date and method
preservation source-file SHA-256 (target) · status · reason if pending
gap false on every figure — mapping is complete; only byte-capture is deferred
Coverage
IG-asset Sourced figures 31
provenance captured 31 (100%)
byte-hashed 0 — pending desktop byte-capture
silent gaps 0
claims-register sources 7 primary releases mapped
Preservation basis: the target hash is the SHA-256 of the exact source file bytes, computed on desktop capture. A fingerprint of retrieved/extracted text is not a substitute and is never presented as the source hash. Kamoa-Kakula and Manono are DRC data-layer records — figures preserved as public-record facts, no editorial position.