Unverified things
stay latent.
A dark room where claims wait undeveloped, and proof is the only thing that gets to be bright.
A gate that fails
hangs red.
45 receipts and 7 seals on one dated wire, every print generated from a gate's own artifact. The failures are not hidden at the dark end. One red print is the control: the check that proves the checker can say no.
36 passed12 mixed04 failed07 sealed13 latent
stop and read the line →the bench · readings
every number derived at build · the command re-derives it
- 01HarnessBenchreads · enforced vs advised1.000EES · n = 68 · tpr 1 · fpr 0as of 2026-08-16
bash scripts/reproduce.shN2 - 02OPERANTreads · judge calibration5judge sittings sealed · 2 pass · 3 fail · T2as of 2026-08-19
npx -y saagar-operant-mcpadjacent - 03CheckSealreads · check receipts6T2 seals in Rekor · 1 T1 · latest rekor 2514456348as of 2026-08-19
checkseal verify-keyless harness-config.sigstore.json --subject harnessbench-semantic-clean-room.config-manifest.jsonN1 - 04Verification Ledgerreads · mechanism vs words4/4guarantees on the served row · record 1 · source_trust agentas of 2026-08-16
python -m verification_ledger.conformanceN4
45 receipts · 7 seals · source 9d06aafd ·walk one artifact down the bench →
newest print · fully fixed
No Sermons in the Seal
The oldest rule in my notes says words bind at about eighty percent and mechanisms at a hundred. This is the methodology note for applying that rule to trust claims themselves: what it took to measure the gap instead of stipulating it, and why the seal that carries the result trusts a digest equality, never a sentence.
Enter the reading room