The problem

Board reporting is usually a slide someone typed.

A signed report whose figures an examiner can compute again is a different kind of document.

How it works, illustrated

Your policy, measured: a policy target over a quarter, and the signed board reportA policy target measured week by week from the signed log: 41 of 42 cases met. The signed board report verifies as PASS; a changed copy fails as NUMBERS DIFFER.Signed logthe only inputdevice_first_seenalert_raiseddecisionconfig_changeThe CLI on a copy of thedatabase and the offlineverifier get the same numbers.Policy IT-07 §4.2 · Q3Decide on new devices within 2 business daysdecide_new_device · weekends and holidays count nothingJulAugSepall meta miss41 of 42 cases metTarget not metJudged on exact counts; the miss links to its log entries.Time to decisionmedian 5 h 12 m90th percentile 1 d 9 hTime to detectmedian 2 m 31 ssighting to alertBoard report2026-Q3Example Community Bankevidence packSigned, with:figures as JSON and CSVone row per casethe evidence packproofs for each entryreport-verify board-report.zipPASSa copy with one figure changedNUMBERS DIFFERYour policy, measured: a policy target over a quarter, and the signed board reportA policy target measured week by week from the signed log: 41 of 42 cases met. The signed board report verifies as PASS; a changed copy fails as NUMBERS DIFFER.Measured from the signed log, and nothing else.Policy IT-07 §4.2 · Q3Decide on new devices within2 business daysdecide_new_deviceJulAugSep41 of 42 metTarget not metOn exact counts. The miss links to its entries.Time to decisionmedian 5 h 12 m90th percentile 1 d 9 hTime to detectmedian 2 m 31 sBoard report2026-Q3Example Community Bankevidence packSigned, with:figures as JSON and CSVone row per casethe evidence packproofs for each entryreport-verify board-report.zipPASSa copy with one figure changedNUMBERS DIFFER
Your policy, measured. Every figure is computed from the signed log, and the report verifier computes it again. Illustration with fictional figures.
What this shows

Every figure comes from one input, the signed audit log: device sightings, alerts, decisions and each saved version of the policy. One policy target, from a fictional policy section IT-07 4.2: decide on new devices within 2 business days. Week by week over the third quarter, every case was decided in time except one, in the seventh week, so 41 of 42 cases met the target and the target is not met; pass or fail is decided on exact counts. Time to decision had a median of 5 hours 12 minutes and a 90th percentile of 1 day 9 hours, and the slowest case links to its log entries. Time to detect had a median of 2 minutes 31 seconds. Beside it, the board report for the quarter, a signed file with the figures as JSON and CSV, one row per case, the evidence pack for the same quarter and the proofs. The report verifier checks the signature and computes every figure again: the signed report prints PASS, and a copy with one figure changed prints NUMBERS DIFFER. All figures are fictional.

appliance report-verify -trust <key-id> board-report.zip# checks the signature, every file, the checkpoint, the proofs and the# evidence pack inside, then recomputes every figure from the log# exit 0: PASS, exit 10: NUMBERS DIFFER
Command-line usage from the appliance’s reference.

The report is print-ready HTML with trends drawn inline, no scripts and its own strict content policy. Print it to PDF from the browser.

The Board report tab is designed and arrives with the new interface. Today the report is generated and downloaded through the data routes.

How it works

  • A signed ZIP: the report, the figures as JSON and CSV, one row per case, the evidence pack for the same period, every earlier entry the figures depend on, and the proofs.
  • The report verifier checks the signature, every file, the checkpoint, every proof and the pack inside, then computes the figures again and requires them to match byte for byte.
  • A number that does not follow from the log fails as NUMBERS DIFFER, even when the file carries a valid signature.
  • The report is bound to its period: the pack’s period must equal the report’s, and the entries at each edge must fall outside it.
  • The figures are your board’s policy targets, measured from the signed log.

What it proves

The board, the examiner and the auditor get one document whose every figure is computed again from the signed log, with the full evidence pack for the period attached.

What it records

Entries this feature writes to the tamper-evident audit log. Each one is in the evidence packs that cover its date.

  • board_report_exportedThe report’s fingerprint, the pack’s fingerprint, the period, entry range, checkpoint, method and targets met.
  • pack_exportedThe evidence pack inside the report.

Limits

What it does not do yet, stated plainly. The technical page lists every known gap.

  • No scheduled emailing yet.
  • PDF comes from the browser’s print dialog.
  • The report cannot prove the absence of state that nothing in the period touched. It says so, and the inventory file in the pack is the cross-check.

Technical detail

Early access: now taking pilot customers.

See it on your own network.

A live demo with the engineers who built it. We bring a simulated network; you bring your questions.