… user-scope path discipline (PR #1081 review threads)

Address 6 unresolved P1/P2 reviewer threads on PR #1081:

1. PRRT_kwDOSF9kNM5-81EK (P2): B-0129 mapped to class #19 was
   wrong — the deferred-per-explicit-policy class is #20, not #19
   (#19 is peer-AI structural / consolidation-prompt). Fixed in
   Composes-with bullet.

2. PRRT_kwDOSF9kNM5-81EL (P2): B-0130 mechanization mapping cited
   "class #4 (wildcard-ref auditor)" but wildcard-not-navigable is
   class #11 in this v2 taxonomy (#4 was a v1-set name with
   different meaning). Fixed to "#11 (wildcard-not-navigable
   auditor)".

3. PRRT_kwDOSF9kNM5-82DA (P1): v1 taxonomy reference in the body
   used in-repo-shaped path `memory/feedback_pr_thread_resolution
   _class_taxonomy_2026_04_28.md` even though labelled user-scope.
   Switched to explicit `~/.claude/projects/<slug>/memory/...`
   form for both occurrences (lines 11 + 17).

4. PRRT_kwDOSF9kNM5-82DY (P1): `.githooks/` example in class #17
   read as describing current mechanism. Reframed as illustrative
   of the structural pattern (versioned-vs-runtime layer); added
   note that this repo currently has no `.githooks/` directory
   and hooks are plugin-shipped via `.claude/settings.json` per
   the hooks audit.

5. PRRT_kwDOSF9kNM5-82Dh (P1): Composite — the class-number fixes
   above + Composes-with cleanup. Removed duplicate v1-parent
   xref line; consolidated into the corrected user-scope-path
   bullet.

6. PRRT_kwDOSF9kNM5-82Dw (P1): Tick-history shard range
   "0904Z.md through 1007Z.md" — `1007Z.md` does exist in current
   tree (added by PR #1080), but reworded to be more robust:
   "from 0904Z.md onward through the 10:00-11:00 UTC drain
   window" — names the directory + the time range without
   pinning a fragile upper-bound filename.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

… user-scope path discipline (PR #1081 review threads)

Address 6 unresolved P1/P2 reviewer threads on PR #1081:

1. PRRT_kwDOSF9kNM5-81EK (P2): B-0129 mapped to class #19 was
   wrong — the deferred-per-explicit-policy class is #20, not #19
   (#19 is peer-AI structural / consolidation-prompt). Fixed in
   Composes-with bullet.

2. PRRT_kwDOSF9kNM5-81EL (P2): B-0130 mechanization mapping cited
   "class #4 (wildcard-ref auditor)" but wildcard-not-navigable is
   class #11 in this v2 taxonomy (#4 was a v1-set name with
   different meaning). Fixed to "#11 (wildcard-not-navigable
   auditor)".

3. PRRT_kwDOSF9kNM5-82DA (P1): v1 taxonomy reference in the body
   used in-repo-shaped path `memory/feedback_pr_thread_resolution
   _class_taxonomy_2026_04_28.md` even though labelled user-scope.
   Switched to explicit `~/.claude/projects/<slug>/memory/...`
   form for both occurrences (lines 11 + 17).

4. PRRT_kwDOSF9kNM5-82DY (P1): `.githooks/` example in class #17
   read as describing current mechanism. Reframed as illustrative
   of the structural pattern (versioned-vs-runtime layer); added
   note that this repo currently has no `.githooks/` directory
   and hooks are plugin-shipped via `.claude/settings.json` per
   the hooks audit.

5. PRRT_kwDOSF9kNM5-82Dh (P1): Composite — the class-number fixes
   above + Composes-with cleanup. Removed duplicate v1-parent
   xref line; consolidated into the corrected user-scope-path
   bullet.

6. PRRT_kwDOSF9kNM5-82Dw (P1): Tick-history shard range
   "0904Z.md through 1007Z.md" — `1007Z.md` does exist in current
   tree (added by PR #1080), but reworded to be more robust:
   "from 0904Z.md onward through the 10:00-11:00 UTC drain
   window" — names the directory + the time range without
   pinning a fragile upper-bound filename.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

… wave + Aaron meta-class (#1081)

* memory(taxonomy v2): consolidate 13 new resolution-classes from 2026-05-01 drain wave + Aaron meta-class

Per Deepseek peer-AI's explicit ask 2026-05-01: *"The review-thread taxonomy
is ready for consolidation. The loop has identified ~13 distinct classes of
review findings."* Plus Aaron's just-in-time meta-flag during the
consolidation: *"that itself is a new class of review"* — observing that
peer-AI structural intervention is itself a distinct class (#19), making
this the consolidation that surfaced its own bootstrapping class.

Extends v1 (`feedback_pr_thread_resolution_class_taxonomy_2026_04_28.md`,
currently user-scope-only — promotion to in-repo is a separate follow-up):

- Phantom-blocker sub-classes 1a/1b/1c: line-prefix-display-artifact /
  rendered-form-misread / hallucinated-content (each requires different
  empirical-verification command).
- Real-fix sub-classes 8-18: forward-reference (9+ instances; canonical),
  contradicts-CLAUDE.md (dual corrective: remove-loophole + cite-principle-
  inline), stale-filename-cross-reference, wildcard-not-navigable,
  executable-prose, persona-name-role-ref, user-scope-only-reference,
  intra-file-drift, named-link, structural-pattern-mismatch,
  same-wake-author-error-cluster.
- New mode classes 19-20: peer-AI-structural / consolidation-prompt
  (the meta-class — how new classes get discovered) and
  deferred-per-explicit-policy / no-edit-resolve (highest-leverage class
  for batch-resolution; cite-and-resolve template scales linearly).

Includes diagnostic flowchart (14-step cascade for first-read classification)
and empirical evidence summary (45+ class-firings catalogued across ~16
ticks). Composes with B-0129 (named-policy that powers class #20),
B-0130 (mechanization path for classes #4/#10/#15/#18), and the
drain-wave tick-history shards 0904Z..1007Z.

Carved candidate: *"Filed-policy beats per-PR judgment. When a recurring
finding has an explicit policy, the cite-and-resolve template scales
linearly across PRs. The 13-PR batch is the worked example."*

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix(taxonomy-v2): correct class-number references + .githooks scope + user-scope path discipline (PR #1081 review threads)

Address 6 unresolved P1/P2 reviewer threads on PR #1081:

1. PRRT_kwDOSF9kNM5-81EK (P2): B-0129 mapped to class #19 was
   wrong — the deferred-per-explicit-policy class is #20, not #19
   (#19 is peer-AI structural / consolidation-prompt). Fixed in
   Composes-with bullet.

2. PRRT_kwDOSF9kNM5-81EL (P2): B-0130 mechanization mapping cited
   "class #4 (wildcard-ref auditor)" but wildcard-not-navigable is
   class #11 in this v2 taxonomy (#4 was a v1-set name with
   different meaning). Fixed to "#11 (wildcard-not-navigable
   auditor)".

3. PRRT_kwDOSF9kNM5-82DA (P1): v1 taxonomy reference in the body
   used in-repo-shaped path `memory/feedback_pr_thread_resolution
   _class_taxonomy_2026_04_28.md` even though labelled user-scope.
   Switched to explicit `~/.claude/projects/<slug>/memory/...`
   form for both occurrences (lines 11 + 17).

4. PRRT_kwDOSF9kNM5-82DY (P1): `.githooks/` example in class #17
   read as describing current mechanism. Reframed as illustrative
   of the structural pattern (versioned-vs-runtime layer); added
   note that this repo currently has no `.githooks/` directory
   and hooks are plugin-shipped via `.claude/settings.json` per
   the hooks audit.

5. PRRT_kwDOSF9kNM5-82Dh (P1): Composite — the class-number fixes
   above + Composes-with cleanup. Removed duplicate v1-parent
   xref line; consolidated into the corrected user-scope-path
   bullet.

6. PRRT_kwDOSF9kNM5-82Dw (P1): Tick-history shard range
   "0904Z.md through 1007Z.md" — `1007Z.md` does exist in current
   tree (added by PR #1080), but reworded to be more robust:
   "from 0904Z.md onward through the 10:00-11:00 UTC drain
   window" — names the directory + the time range without
   pinning a fragile upper-bound filename.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>