fix(tick-shard-0335Z): escape literal pipes inside inline code (#1297 P2 follow-up)

docs/hygiene-history/ticks/2026/05/03/0335Z.md

@@ -1 +1 @@
-| 2026-05-03T03:35:00Z | opus-4-7 / autonomous-loop continuation | a2e2cc3a | **Thread-resolution cluster across 5 PRs (8 findings; 2 substantive fixes + 6 stale/convention-misread).** Cycle worked: refresh revealed 5 in-flight PRs all blocked on threads + checks. Investigated each: #1291 (cross-disciplinary memo) — 2 P2 duplicate MEMORY.md entries (REAL find from rebase auto-merge merging verbose entries with trimmed entries); fixed via dedupe (lines 12, 14-17 removed). #1293 (tick shard 0320Z) — P0 schema mismatch (STALE; diff-view-line-number-misread; hexdump confirmed file starts with `\\x7c \\x20 \\x32` = `| 2`). #1294 (fragment phrasing) — P2 length tightening (REAL; entry tightened from 430→313 chars). #1295 (edge-defining-work memo) — P1 markdown-table-double-pipe (STALE; awk + grep confirmed table is `^| Tier | What it is |` single-leading-pipe). #1296 (guess #003 — already MERGED) — P1 convention-misread; individual guess files don't need MEMORY.md entries because directory README is indexed. **Pattern observation**: of 8 findings, 2 were real (~25%); 6 were stale or convention-misreads (~75%). The stale-finding rate is HIGH for in-flight cluster reviews. The 2 real fixes (dedupe + length-tighten) were both about MEMORY.md hygiene that compounded over the calibration cluster. **Substrate-claim-checker class observation**: rebase-auto-merge-creates-duplicates is a new failure mode worth cataloguing — when chained PRs and main both edit MEMORY.md, the rebase merge can put both versions in the file. B-0170 v1+ semantic-equivalence sub-class would catch this (two entries pointing at the same memory file = semantic-equivalence drift). | #1294 (fragment phrasing tightened) wait-ci, auto-merge armed; #1295 (edge-defining memo) wait-ci with stale thread resolved; #1293 (tick 0320Z) wait-ci with stale thread resolved; #1291 (cross-disciplinary memo, deduped) wait-ci, auto-merge armed; #1296 (guess #003 + recovery) MERGED with convention-misread thread resolved | This tick teaches **stale-finding rate as immune-system signal**: when a cluster of in-flight PRs generates 75% stale findings, it suggests reviewer-tooling-context-lag (reviewers see PR-branch state, not main state, after rapid sequential merges). This is the review-against-PR-branch-not-main class operating at scale — not 1-2 instances per PR but multiple-instances-per-PR-cluster. The discipline: investigate each finding (don't auto-resolve), but expect high stale-rate during rapid-cluster-merge windows. The remaining-real-rate (~25%) is the actionable substrate-quality data. |
+| 2026-05-03T03:35:00Z | opus-4-7 / autonomous-loop continuation | a2e2cc3a | **Thread-resolution cluster across 5 PRs (8 findings; 2 substantive fixes + 6 stale/convention-misread).** Cycle worked: refresh revealed 5 in-flight PRs all blocked on threads + checks. Investigated each: #1291 (cross-disciplinary memo) — 2 P2 duplicate MEMORY.md entries (REAL find from rebase auto-merge merging verbose entries with trimmed entries); fixed via dedupe (lines 12, 14-17 removed). #1293 (tick shard 0320Z) — P0 schema mismatch (STALE; diff-view-line-number-misread; hexdump confirmed file starts with `\\x7c \\x20 \\x32` = `\| 2`). #1294 (fragment phrasing) — P2 length tightening (REAL; entry tightened from 430→313 chars). #1295 (edge-defining-work memo) — P1 markdown-table-double-pipe (STALE; awk + grep confirmed table is `^\| Tier \| What it is \|` single-leading-pipe). #1296 (guess #003 — already MERGED) — P1 convention-misread; individual guess files don't need MEMORY.md entries because directory README is indexed. **Pattern observation**: of 8 findings, 2 were real (~25%); 6 were stale or convention-misreads (~75%). The stale-finding rate is HIGH for in-flight cluster reviews. The 2 real fixes (dedupe + length-tighten) were both about MEMORY.md hygiene that compounded over the calibration cluster. **Substrate-claim-checker class observation**: rebase-auto-merge-creates-duplicates is a new failure mode worth cataloguing — when chained PRs and main both edit MEMORY.md, the rebase merge can put both versions in the file. B-0170 v1+ semantic-equivalence sub-class would catch this (two entries pointing at the same memory file = semantic-equivalence drift). | #1294 (fragment phrasing tightened) wait-ci, auto-merge armed; #1295 (edge-defining memo) wait-ci with stale thread resolved; #1293 (tick 0320Z) wait-ci with stale thread resolved; #1291 (cross-disciplinary memo, deduped) wait-ci, auto-merge armed; #1296 (guess #003 + recovery) MERGED with convention-misread thread resolved | This tick teaches **stale-finding rate as immune-system signal**: when a cluster of in-flight PRs generates 75% stale findings, it suggests reviewer-tooling-context-lag (reviewers see PR-branch state, not main state, after rapid sequential merges). This is the review-against-PR-branch-not-main class operating at scale — not 1-2 instances per PR but multiple-instances-per-PR-cluster. The discipline: investigate each finding (don't auto-resolve), but expect high stale-rate during rapid-cluster-merge windows. The remaining-real-rate (~25%) is the actionable substrate-quality data. |