All eight stages, a clean tree. loops() on every boot: heat 6, 1000 cooled to 0, variance not 0 and width 455 with L8 held on card 18, 66 card changes ending on card 11 -- the same on amd64, aarch64, riscv64 and the stick. dict_hash 0x7c5a2f0c02145e63, POST 992/992. The record is logs/20261120-192820/README.md. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017RzZgrf7V7Zk2iczBXDRbm
The protocol on 65737e28: PASS (2026-11-20)
PROTOCOL.md here: all eight stages passed on a clean tree (the four acceptance boots, test-32, the
proofs, the images, the documents). It is the evidence for the PR after #198.
The loops, asserted on every boot (#189, ENGINE step 4 Phase 7, Task 9)
scripts/qemu_session.py's loops(), at the prompt of each v4 boot, gave the same values on all four:
| Check | amd64 | aarch64 | riscv64 | stick-high |
|---|---|---|---|---|
| L1, heat: PHY1 called six times from code | H=6 | 6 | 6 | 6 |
| L3, decay: PHY1 set to 1000, left alone 40 s | D=0 | 0 | 0 | 0 |
| L2 and L5, L8 held on card 18: variance not 0, width | -1, 455 | -1, 455 | -1, 455 | -1, 455 |
| L8: changes, card | 66, 11 | 66, 11 | 66, 11 | 66, 11 |
The same on every architecture, as one heartbeat and integer arithmetic should give. dict_hash
0x7c5a2f0c02145e63 on all three, and POST 992/992 on all four boots (floor 992).
The run before it
The protocol on a9a8687f (logs/20261120-184418/PROTOCOL.md) FAILED at its last stage, documents:
docs/reference/v4-dictionary.tsv had no entries for L8-CARD@ and L8-HOLD, the Task 9 words. That
was an omission in the work, not in the gate. Its other seven stages passed, with the values above.
65737e28 added the two entries, and the protocol was run again from the start.
What is in it, since #198
- ADR step 3, task 4: the host placer (
v4/src/place.c) and a small train placed on Hera at her HERE, with an import from the nucleus. - fe25519: the carry taken out by multiplying, not by a left shift of a negative value; the nightly's
sanitizer failure of 2026-11-20, test first (
v4/tests/test_fe25519.c). - #189 Task 9:
L8-CARD@andL8-HOLD, the loops asserted at every boot,WORD_PHYSICS; POST 992. capsules/BLOCK_MAP.mdrefreshed; CLAUDE.md, "while a PR is open, commit locally and push once" (#152).