session 2026-09-04: PENDING-179 gates, OWED-5 queued, Active Session rotated

Session record, ledger merge, MEMORY.md rotation (prior Active Session
demoted verbatim to MEMORY-reference.md), 6 KG triples, and OWED-5 queued
in PENDING-141's ratified owed-entries list.

OWED-5: a must-detect control must report its denominator, and a denominator
of zero is a FAIL. Steward-stated 2026-09-04, queued unruled — the ladder is
frozen under REVIEWED-123, and turning the rule on converts currently-green
controls to red across the fleet, which is a ruling rather than an edit.

Earned on the vacuous PASS (0/0) that nearly certified a broken discriminator.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01X3L79vgAnt1x2kxvf23Qt7

claude/memory/knowledge-graph.jsonl

@@ -773,3 +773,9 @@
 {"subject": "claude-code", "predicate": "drift-pattern-good-direction", "object": "DIAGNOSED BEFORE PRUNING, AND THE DIAGNOSIS INVERTED THE ACTION. Asked to fix MEMORY.md's load-budget breach, the obvious move was to cut the longest entries. A per-line byte map plus a duplication check found instead that the index had become the SOLE CUSTODIAN of live state for three workstreams, and that the longest entries in 'Rules that fire silently' are long BY DESIGN. Pruning would have deleted three undocumented workstreams and removed the catches that fire when I would not know to look them up.", "valid_from": "2026-09-03", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-03-the-index-had-become-the-record.md", "extracted_at": "2026-09-03"}
 {"subject": "the discipline 'prove the instrument before trusting a clean line'", "predicate": "prevention", "object": "CAUGHT A PERMANENTLY-FAILING CONTROL BEHIND A CLEAN REPORT, AND FOUND PENDING-178 AT A SECOND SITE. wake-digest.py reported 407 pointers / 0 dead; its --selftest reported 86 of 87 with overall FAIL. The failing control samples the 13 most recent transcripts and ALL 13 are Tarbuckle mumbles (~66 KB, one human turn each) — the mumble pollution filed as PENDING-178 against the ladder counter has also displaced real sessions out of the digest's own end-to-end control, which will now fail on every run. A lesson banked about instruments stopped a different failure class: alarm-decay, not a false alarm.", "valid_from": "2026-09-03", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-03-the-index-had-become-the-record.md", "extracted_at": "2026-09-03"}
 {"subject": "MEMORY.md", "predicate": "failure-mode", "object": "AN INDEX BECOMES THE RECORD WHEN ITS TRACKER GOES QUIET. Observed 2026-09-03: the studium tracker's chronological log stopped 2026-08-13 while the MEMORY.md one-line pointer carried engine state to 2026-09-01; the Fool's only link was a SEALED seed with nowhere to append; L1's replay mechanism and completion criterion existed in the index alone. The tell is a tracker going quiet while its one-line pointer grows. The rotation at wrap handles the Active Session block; nothing handles slow growth in the tracker list, so the mechanism that produced this breach is unfixed.", "valid_from": "2026-09-03", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-03-the-index-had-become-the-record.md", "extracted_at": "2026-09-03"}
+{"subject": "claude-code", "predicate": "drift-pattern", "object": "A MUST-DETECT CONTROL WENT GREEN ON AN EMPTY DENOMINATOR. Building ground truth for the mumble-discriminator gate, the signature was copied from tarbuckle-invoke.py -- the file just read -- and matched 0 of 65 transcripts, because mumbles are written by three OTHER fool surfaces. The gate printed 'PASS (0/0)'. Accepting it would have certified a broken discriminator and wired it into three more sites. Caught by 'a null search is evidence about the QUERY', a human-held rule, not by the instrument. Now queued as OWED-5: a must-detect must report its denominator and a denominator of zero is a FAIL.", "valid_from": "2026-09-04", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-04-the-discriminator-was-a-coincidence.md", "extracted_at": "2026-09-04"}
+{"subject": "claude-code", "predicate": "drift-pattern", "object": "FRAME-INHERITANCE ACROSS A QUESTION CHANGE, and it was the session's whole finding. human_turns() carries three passing controls, all testing the question it was BUILT for ('did an executor run unattended?'). The repair plan reused it for a different question ('is this a mumble?') and inherited the controls' authority across that gap. Re-run at the scope of the extension: must-detect 22/24, must-not-flag 35/41. The exclusion works only when a mumble happens to quote a slash command; the 2 leaks are exactly the 2 marker-free mumbles.", "valid_from": "2026-09-04", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-04-the-discriminator-was-a-coincidence.md", "extracted_at": "2026-09-04"}
+{"subject": "claude-code", "predicate": "drift-pattern-good-direction", "object": "STOPPED AT THE GATE AND DID NOT BUILD THE REPLACEMENT. When 2a failed, the handover's stated reason was honoured rather than its instruction alone: three controls passed and a fourth broke, so the finding is about the CONTROL SET, and a successor discriminator written under the momentum of the break inherits whatever made the first set look sufficient. Sites 2-4 left unrepaired on purpose. Gates 2b and 2c were still run, because 2a does not gate them.", "valid_from": "2026-09-04", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-04-the-discriminator-was-a-coincidence.md", "extracted_at": "2026-09-04"}
+{"subject": "the rule 'a null search is evidence about the QUERY'", "predicate": "prevention", "object": "STOPPED A VACUOUS PASS FROM CERTIFYING A BROKEN DISCRIMINATOR, 2026-09-03. The rule was banked 2026-08-31 after four name-search errors in two sessions -- a different failure class entirely (reporting 'you don't have X' from a filename search). Here it fired on a control's empty denominator and prevented a broken predicate being wired into three further consumer sites. A lesson banked from one class stopping another is the transfer signature.", "valid_from": "2026-09-04", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-04-the-discriminator-was-a-coincidence.md", "extracted_at": "2026-09-04"}
+{"subject": "~/_Dev/claude-transcript-archive", "predicate": "is-not-version-controlled", "object": "NO .git, no parent repo, 144 MB, single copy on one disk -- while preserve-transcripts.py:11 states it copies transcripts 'to a git-tracked location so the population stops shrinking'. 11 transcripts have been pruned at source and exist ONLY here. PENDING-144's class (substrate claims inside governance scripts checked by nothing) at the site where it costs most, because the docstring is what a reader consults to decide whether the evidence is safe. Steward directed a git init as the next session's first act.", "valid_from": "2026-09-04", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-04-the-discriminator-was-a-coincidence.md", "extracted_at": "2026-09-04"}
+{"subject": "the ladder trial counter", "predicate": "composition", "object": "N-now 65 of 84 measured 2026-09-03 = 41 real sessions + 24 Tarbuckle mumbles (36.9% machine chatter). Rising fast. The prior record in MEMORY.md said 44 and falling ('shedding faster than it gains') -- wrong in the number and backwards in the direction; corrected record-only. Also: preservation has permanently diverged the two stores (11 archive-only files), so grading must now specify WHICH STORE, not only what N -- the trigger reads live, honest grading of a post-08-07 population must read preserved.", "valid_from": "2026-09-04", "valid_to": null, "confidence": 1.0, "source_file": "session-2026-09-04-the-discriminator-was-a-coincidence.md", "extracted_at": "2026-09-04"}