[danielgorgonha]

What this changes and why:

Fixes, before the 17 November 2026 verdict runs, the choices an operator could otherwise make on the day:

• Ranges. T is the highest block whose block time is before 2026-11-17 00:00:00 UTC, with every block above it at or after the cutoff. Daily = T-3,455..T, first scan = T-49,999..T, and the first-scan-43d information row = T-149,999..T (150,000 blocks). Anyone can check T from block times on an explorer.
• Browser and laptop. The verdict row on the Galaxy A72 is Firefox. The laptop proof verdict row is Chrome on a Dell G7, 24 GB. Other browsers are reported as information.
• Endpoint. The verdict row goes through https://m1-rocks.konclave.xyz; https://m1-global.konclave.xyz is reported as information. The docs no longer say "the one endpoint".
• Session. For each verdict row, the first session started on 17 November 2026 is the verdict, whatever it gives (pass, FAIL or short). No row is re-run to get another number.
• scripts/fix-ranges.mjs. Finds T from compact block times (GetLatestBlock, then bisection and a forward walk, since block times are not monotonic) and prints T, its time and hash, and the three ranges. With --write it writes only the six heights into web/ranges.mjs. It needs --read-approved and refuses a tip before the cutoff or fewer than 100 blocks past T.

The summary code is unchanged: it prints a pass or a fail for any row that clears the rules, and the protocol names the verdict row.

Tests run (see CONTRIBUTING.md):

node --test tests/node.test.mjs tests/charts.test.mjs: all pass. The fix-ranges tests run against the repo's mock chain and cover the boundary (a time equal to the cutoff, and one second before), a non-monotonic dip, the tip refusals, the refusal without --read-approved, and --write changing only the heights. Each was checked to fail with its logic removed.

For a result under community-results/: commit built, device, browser:

Not applicable.

AI disclosure: no AI assistance / AI-assisted (tool and what it did):

AI-assisted: Claude Code wrote the change, the tests and the docs; a separate AI review pass audited them. The decisions are the maintainer's.

🤖 Generated with Claude Code