Optional HARD tier
TEE attestation
Hardware environment and bound measurements. Requires the independent fallback.
A quote alone does not establish which inference ran. This tool does not verify quotes.
What proofs are miners submitting and what do they cover?
flop proofs --helpRecorded evidence / devnet
The public proof rail and paid compute channels expose different evidence. Follow each claim to its source before treating a settlement as verified.
pnpm --filter @flop-lab/proofs build && node packages/proofs/dist/cli.js listDownload raw recording ↓Specification / section 3
The line tracks your reading position. It does not show verification progress. Requirements below come from the pinned paper.
Optional HARD tier
Hardware environment and bound measurements. Requires the independent fallback.
A quote alone does not establish which inference ran. This tool does not verify quotes.
Required for every session
Commitment to prefill activations for later checks.
Detection is conditional on validated cells and the commitment, challenge and adjudication assumptions. Decode coverage is open.
Required fallback
Sampled or challenged work is checked in a different trust domain.
Deterrence depends on effective monitoring, available evidence and collectible penalties. A sample does not check every turn.
Direct-rail credit gate
Validator attestations bind the credited statement. Merkle aggregation supports dispatch.
Aggregation is not a proof of the inference forward pass. The indexer does not expose the validator bundle.
Separate payment lane
Direct PoUI credit covers synthetic, ghost, calibration and simulator work. A proof row does not cover a paid session just because the miner matches.
paid agent sessions settle through the compute channel (§12.1)Pinned paper, line 636 ↗
Observation / direct rail
A successful API read can return an empty table. No synthetic submissions are shown here.
pouiProofs
This does not establish that a paid channel lacked its required evidence.
The proof query ended on a short page. It exposes no total.
Coverage / paid channels
The public schema cannot bind a proof to a paid channel. UNVERIFIABLE describes this evidence limit, not a failed proof.
349 of 349 recorded settlements are UNVERIFIABLE.
All rows returned by reported pagination. Reported total: 349.
349 matching settlements. Showing 1 to 12.
0x90f8e2ae1fc38b081c7cd3f58f5892a1720e4b5970bdc88bf234ee8829d10574Block 3,004,365 / 2026-11-06 01:11:24.000 UTC
0xc7a34b3a18a48bd6c7c75aeecb2d7d6909984ef25d2c001d8c2cdbe49decb4e2Block 3,001,399 / 2026-11-06 00:21:58.000 UTC
0x41debee410f0fabe29f8ca8b2276c887ed1bc3c80ed18235f1512d4445cb9060Block 2,999,883 / 2026-11-05 23:56:42.000 UTC
0x2770860866edc09925865d091847ea4252517adbbb7043a1678e7d2f340961acBlock 2,997,258 / 2026-11-05 23:12:54.000 UTC
0x67a3adb51536ef184f03b1653b860d1e667b35d2b6cb7fbc62cd089142319809Block 2,997,176 / 2026-11-05 23:11:32.000 UTC
0xdad6afc3f2a8914a45c41a05cec94fff002a3d6099949bcc02438f399c669dc8Block 2,994,492 / 2026-11-05 22:26:44.001 UTC
0xeda44d4053757bf88f7147f16daf2576d4c6550e560dd10a9a524a3bd664ec8eBlock 2,993,105 / 2026-11-05 22:03:34.000 UTC
0x5ed433a88aa4cbfd7c082a4fb63928f2e3e4df821208f7054c91b9279c52e0f3Block 2,987,921 / 2026-11-05 20:36:59.000 UTC
0x7dd8b561edf99badb64145f21fbdf45168d3aa2783b35631c109cc2460a5faf8Block 2,980,304 / 2026-11-05 18:29:55.000 UTC
0x68c98715d37d137e21df845f764347d8de631e6c5fceb26be0cef836aa318460Block 2,980,169 / 2026-11-05 18:27:40.000 UTC
0xcf00ede2be1863b2c26f0c32500bedcec74b23fd8e57570379a8bdc2fa97d05fBlock 2,979,893 / 2026-11-05 18:23:04.000 UTC
0xb2dc4989e811e9a9e4930acc16181a1ed062c464b239921c7ef2a7e20bb8c4daBlock 2,964,740 / 2026-11-05 14:09:34.000 UTC
Recorder / separate observations
The bundled recorder runs are close together. They do not establish a daily trend. Missing counts stay missing.
0 reported proofs24h
0 sampled leaderboard rows
0 reported proofs24h
0 sampled leaderboard rows
Rolling proofs24h windows overlap. Never sum them. Leaderboard row count is the recorder sample size, not the number of submitted proofs. Open recorder history ↗
Specification only / sampled audit
These defaults are quoted from Appendix A. This tool does not read deployed runtime storage or calculate an empirical detection probability.
sampled_audit_alpha_ppm50_000 ppmAppendix A, line 2804 ↗Default fraction for post-epoch sampled-audit selection.
audit_quantum_gn100_000_000 G_nAppendix A, line 2806 ↗Served-work quantum that forces an audit ticket.
high_value_gn_threshold10_000_000 G_nAppendix A, line 2807 ↗A turn at or above this work threshold is force-audited.
Paper 2a2cc821 fetched 2026-11-06 01:30:01.928 UTC. Read the pinned source ↗
What this does not verify
Sequential public indexer observations. Completion means reported pagination only, not atomic chain state. No cryptographic proof verification or current network claim.
No signatures, hardware quotes, TOPLOC openings, validator bundles, inference execution or runtime parameter enforcement are checked. No wallet or write transaction is used.