Match the observation
Every result links to a received time and source. Different heights or settlement revisions are not treated as equal snapshots.
Check the totals. Follow the disagreement.
Expand any rule to see its values, source responses and arithmetic. Agreement here is indexer consistency, not independent chain proof.
Recorded public data
Download evidence ↓Pass19
The sampled values agree.
Fail1
A declared rule does not hold.
Unverifiable8
The inputs cannot establish a result.
10 HTTP requests · 4,496 ms observed · row limit 3 · no live calls from this page
Invariant board
Start with what disagrees.
Failures first. Unknowns stay visible.
Showing 28 of 28 checks from this recording.
How to read this audit
A result is only as strong as its inputs.
The line tracks your place in this explanation.
Read the exact rule
The stake check defines total as all bonded parts. A label mismatch can fail that rule without proving a runtime fault. Raw planck amounts keep every digit.
Keep the unknowns
Public miner data omits calibration acceptance and expiry blocks. A partial page cannot establish total supply or all inference payments. UNVERIFIABLE never means zero.
Run your own observation
A bounded audit. An inspectable result.
Build the suite from the repository, then run the command. The page is a dated recording. Your CLI run collects a new sample.
flop indexer run --limit 20 --jsonWhat this does not verify
MCP and explorer can share a backend. This board does not verify node storage, cryptographic proofs or inference. It does not claim that existing reports #75 or #76 are fixed. Sequential observations can span a head change and devnet can reset.