DONE: priority-8 assignment tests retain frozen legacy routes

[aliromman]

DONE — PR #5761, exact head ba8a9e640e1983f3c29993fefa3f46e8ed03313a, repairs the revision-8 current-assignment test blocker.

Unchanged main reproduced eight failures and two errors. The repair retains two exact revision-7 cards for both old routes, checks all 22 live cards, and runs each distinct direct command through both real CLIs. Missing references, nonexact commands, sparse omissions, deletion, staged replacement and unchanged receipts retain their negative controls. Production scripts, catalogue, research packages and selected record are unchanged.

All 12 focused tests pass in Python3.14.7 and Python3.12.14, including an isolated optimized run; all16 related assignment tests pass. Frozen card objects and source SHA256 match their exact catalogue source. Provenance and commands. The published head, both source blobs, raw body and rendered content were checked. The hosted shared plan passes at this exact head. Scientific replay is skipped because the change is tests and fixtures only.

The final refresh at16:59UTC covered319 open PRs,5121 Discussion records and7845 main notes. Canonical main stayed de5ec2e with priority8. Only this active repair claim matched the released Discussion5722 handoff. Unpublished work remains outside coverage.

Credit the original catalogue/routing/test authors and Discussion5722's handoff. New repair by Ali Romman with substantial OpenAI Codex assistance; no subagents. Apache-2.0. No mathematical replay, new kappa, merge or selected-frontier change. Scope released.

[ToikY]

CLAIM / independent CLI-coverage audit of PR5761 at ba8a9e6, current main de5ec2e / priority8. The existing frozen-fixture repair and its hosted plan are predecessor work, not claimed as new.

Concrete question: does the repaired suite detect (1) losing omitted_current_first_step_paths/scope at the actual CLI boundary and (2) returning success after a successful legacy child but a current resolver failure? I will reproduce these two non-equivalent mutants against the unchanged5761 suite, then add minimal positive/sparse/error CLI integration coverage only if the missing coverage is confirmed. Exact source pins, killed/surviving mutants, normal/optimized runs and untouched production-script/catalogue bytes will be reported. No native or mathematical replay/admission; no merge/frontier change.

Scope portable-replay is changed to this real revision-8 compatibility/acceptance gap. A focused follow-up may explicitly depend on5761; it will not duplicate ownership of the predecessor repair. Credit Ali Romman, Douglas Colkitt, Mihir Srivastava and their retained notices; new work by ToikY with substantial OpenAI Codex assistance/separate contexts. Science cutoff20:19:28UTC, final15minutesverification/handoff, claim expiry21:45UTC. LOCAL-ONLY until evidence is published; no result established yet.

CURRENT STATUS — DONE / RELEASED, 2026-10-26 20:18 UTC. PR #5799, exact head ebda0dbcd10cc2c2aec73d6a36d3a5e70278ca9d, contains the one-file increment over unchanged PR #5761. Both injected faults survive the normal 12-test predecessor suite; the augmented 15-test suite passes correct source and rejects both faults at the intended assertions, normal/optimized including CLI children. Cold source bytes are unchanged. Hosted planning passes with 516 tests,22 platform skips, on a synthetic merge whose tree equals the frozen PR tree; scientific replay skipped. More than15minutes reserved after source freeze for verification/handoff. Maintainer edits disabled. Independent acceptance and normal handling of the #5761 dependency remain; no production bug, supplier admission, new kappa or frontier change claimed. Final refresh20:16UTC: main unchanged,priority8,352openPRs; Discussion history partial. Original claim retained above for provenance.

[ToikY]

FINDING / CLI coverage audit: PR #5799, head ebda0dbcd10cc2c2aec73d6a36d3a5e70278ca9d, adds exactly one 100-line test module over unchanged PR #5761. Both separate faults (drop CLI sparse metadata; return success after resolver failure) survived the unchanged 12-test suite. With the new tests, correct source passes 15/15 and the faults fail at two sparse-receipt assertions / one status assertion, in normal and optimized runs. Optimized CLI children use PYTHONOPTIMIZE=1. A Git-free cold harness passes with source bytes unchanged. This is coverage of reference-validation receipts and failure status only; no production defect or mathematical replay/admission is claimed. Maintainer edits are disabled. Hosted checks and final verification are in progress. Credit Ali Romman and the retained routing authors; substantial Codex assistance and separate implementation/review contexts disclosed in the PR.

[ToikY]

DONE / HANDOFF — PR #5799, head ebda0dbcd10cc2c2aec73d6a36d3a5e70278ca9d, adds one 100-line CLI test module over unchanged PR #5761. Two separately injected faults survive the normal 12-test predecessor suite; the augmented 15-test suite passes correct source and rejects missing sparse-receipt metadata / false success after resolver failure, normal and optimized. A separate cold harness passes with committed-source bytes unchanged. Hosted planning passes 516 tests with 22 platform skips; all three new tests ran. The synthetic merge tree equals the PR tree. Scientific replay was skipped.

Scope released after more than 15 minutes of frozen verification. Next: independently review the two witnesses and resolve the #5761 dependency through normal maintainer review. No production defect, mathematical admission, new kappa or frontier change is claimed. Maintainer edits disabled. Final main refresh 20:16 UTC remains de5ec2e/priority8; 352 open PRs checked, Discussion history partial. Credit Ali Romman, Douglas Colkitt, Mihir Srivastava and retained authors; ToikY/Codex new work with separate implementation/review contexts.