Compare each pinned clause with what the public devnet exposes. See agreement, missing evidence and the exact quoted sides.
Recorded comparison Last compared
01 / Result map
Every result has a boundary.
393 compared rows. Select a result to inspect it.
02 / Evidence path
Read the claim. Inspect its reach.
The line tracks reading progress.
Pin the paper
Appendices A, G and H retain line anchors. Public vectors retain their declared status and expectations.
Quote the observation
Catalog values and 100 sampled top-level transactions carry their own reception time.
Keep the unknown
Missing activity does not prove missing code. Signature and implementation checks need independent runtime evidence.
03 / Clause explorer
Open the two sides.
393 rows match these filters. Expand any row for evidence.
A
Parameters
Enforced values, reference values and summary clauses against the public catalog.
200 rows
G
Extrinsic index
A bounded top-level call sample. Positional values do not reveal argument names.
66 rows
H
Implementation status
Informative claims from the paper. An empty activity count cannot verify a mechanism.
76 rows
V
Located vectors
Public cases, pending drafts and wire families. Located, not executed.
51 rows
04 / Reproduce
Take the evidence with you.
From a checkout of flop-lab with dependencies installed. The default replays the recorded data. Add --live for a bounded fresh MCP comparison against the same paper pin.
pnpm --filter @flop-lab/spec build && node packages/spec/dist/cli.js diff --section A
Deployed storage constants, runtime metadata, finalized inclusion, vector execution and implementation-status claims. Missing calls in a bounded sample are not absent from the chain. No write transactions are made.