Method — Provenance
Public statement of how every claim in this knowledge base is sourced and verified. (Machine counterpart: the claim registry.)
Tags
| Tag | Meaning | Normative? |
|---|---|---|
[src] |
Recovered or blockscout-verified Solidity, read directly from the deployed bytecode’s verified source | yes |
[chain] |
Onchain query (eth_call / eth_getLogs / receipts / storage) at a stated block | yes |
[ext] |
Community material (sites, spreadsheets, trackers) — discovery-grade only | no |
[irc] |
Verbatim line-referenced quote from the #atropa_logged corpus (2023-10-20 → 2024-12-03) | no |
[ext-owner] |
Owner-provided, unverified | no |
Normative statements must carry a proof pointer (address + tx hash / eth_call selector / source file + lines) and, where state-dependent, a pinned block. Corrections are silent supersessions: the public text is current-state, and the supersession chain lives in a versioned research log (the project’s provenance back-end, outside this published tree).
Claim registry
The claim registry is generated from the same data layer as the pages and
holds claim-class rows: {id, statement, provenance, proof_ref, pinned_block, artifact}. A page statement and a registry row must never disagree; the
registry is the machine-checkable half of the contract.
Reproducibility
Every exported data artifact carries a manifest and a pinned block; the reference bundles under /export are the exact files this portal renders from. The taxonomy regenerates from the merged census with a self-test that re-derives class counts and fails on any mismatch.