Skip to content

audit: 2026-09-17 07:00Z — T-2 pass 3, the 602nd site removed (#3415), rulings applied, T-4 stops at the dry-run #3417

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
noahgift merged 1 commit into from
Sep 17, 2026
Merged

audit: 2026-09-17 07:00Z — T-2 pass 3, the 602nd site removed (#3415), rulings applied, T-4 stops at the dry-run #3417

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions docs/audits/autonomous-run-2026-09-15.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -647,3 +647,15 @@ scheduling chance, and it costs ~3× (gx10 10.6 min vs intel 34.5 for step 1).
### Still open

- #3412 → #3411 → T-1 → T-2 → T-3 (clean-room on the tag) → T-4 → close → ledger record.

## Interval — 07:00Z (2026-09-17) — pass 3, one row; rulings applied

### What moved

- **T-2 pass 3 on `5d95ed54e` (06:18Z): NO-GO on ONE row** — `pmat-comply` CB-200 602 vs 601. Roster diff (pmat's own comply index, release tree vs the CB-200 fix tree): the single addition is `crates/apr-cli/src/commands/parity_03.rs::run`, taken from B to B- (cyc 11) by #3412's refusal call site. **#3415** extracts `refuse_unroutable` + `head_geometry` (both A+); `run` → B+ (cyc 7); the tree's sub-B count is 601 with both baseline lines untouched at 601 — **the 602nd site is removed, nothing restamped** (operator ruling 1, verified before merge). Every other T-2 row was green on pass 3: `pv-contracts` 1841/1841, `model-parity` reports `UNMEASURED-TOOL` for the MoE/qwen35 rows, perf041 marker fresh.
- **Operator rulings 2026-09-17, applied**: (2) §4 T-0 now says the T-2 preflight runs on `main` HEAD *before* the bump PR opens, the bump PR is refused at arm time without a GO receipt for its parent sha, and the release commit moves zero times — from 0.69; (3) §4 T-2 + §7: a nightly feeding a T-2 row that fails ≥2 consecutive runs auto-opens an issue on the next milestone and reports `NIGHTLY-RED` — mechanism ticket **#3416** (0.69); first entry: `cuda-nightly.yml` 5 consecutive fails → #3096 (stays 0.70); the ledger record now carries the PP-26 witness provenance (lambda sm_89) beside the red producer. **T-4 for this train is the operator's**: the driver is re-pointed to stop after the cascade dry-run receipt (`to-step=dryrun`, pid 2553564).
- Merged: #3410 05:13Z, #3412 05:44Z, #3411 05:44Z, #3414 06:23Z.

### Still open

- #3415 (queued, group on intel) → T-1 → T-2 pass 4 → T-3 (tag, clean-room dispatched on the tag with HEAD==tag asserted, `apr-*` assets, install.sh rows on intel + gx10) → T-4 dry-run receipt → STOP and report.
Loading