Lucent-Financial-Group · AceHack · May 21, 2026 · May 21, 2026
diff --git a/docs/pr-discussions/PR-4471-docs-shadow-maji-anti-entropy-log-on-blob-slop-prs.md b/docs/pr-discussions/PR-4471-docs-shadow-maji-anti-entropy-log-on-blob-slop-prs.md
@@ -0,0 +1,57 @@
+---
+pr_number: 4471
+title: "docs(shadow): Maji anti-entropy log on blob slop PRs"
+author: "AceHack"
+state: "MERGED"
+created_at: "2026-05-21T01:49:45Z"
+merged_at: "2026-05-21T02:15:19Z"
+closed_at: "2026-05-21T02:15:19Z"
+head_ref: "lior/shadow-drift-blob-slop-4467"
+base_ref: "main"
+archived_at: "2026-05-21T02:43:39Z"
+archive_tool: "tools/pr-preservation/archive-pr.ts"
+---
+
+# PR #4471: docs(shadow): Maji anti-entropy log on blob slop PRs
+
+## PR description
+
+Shadow lesson log documenting the high-entropy semantic slop in PRs 4467 and 4466 masquerading as atomic decomposition.
+
+## Reviews
+
+### COMMENTED — @copilot-pull-request-reviewer (2026-05-21T01:52:09Z)
+
+## Pull request overview
+
+Adds a research “shadow lesson log” documenting an incident where very large, multi-domain pull requests were mislabeled as “atomic decomposition,” and captures the intended operational lesson (avoid accidentally committing unrelated working-tree changes).
+
+**Changes:**
+- Introduces a new research note summarizing the observation, critique, actions taken, and lesson learned from PRs #4467/#4466.
+- Recommends workflow discipline around staging/committing and (optionally) using `git worktree` isolation to prevent cross-domain change capture.
+
+## Review threads
+
+### Thread 1: docs/research/2026-05-21-maji-shadow-lesson-log-blob-slop.md:15 (resolved)
+
+**@copilot-pull-request-reviewer** (2026-05-21T01:52:08Z):
+
+P1: The referenced tick file path `docs/hygiene-history/ticks/2026/05/19/1614Z-c.md` does not exist in the repository tree, so this cross-reference will be broken/misleading for readers. Please update this to the correct existing path (or reference the PR/commit where the file lives, if it was only present in PR #4470).
+
+### Thread 2: docs/research/2026-05-21-maji-shadow-lesson-log-blob-slop.md:17 (resolved)
+
+**@copilot-pull-request-reviewer** (2026-05-21T01:52:08Z):
+
+P2: "Git locks are clear" is ambiguous and easy to misread as a runtime/process-held lock. Consider using more specific terminology (e.g., no `.git/index.lock` present, or no locked worktrees per `git worktree list --porcelain`) to avoid repeating the known lock/metadata confusion.
+
+### Thread 3: docs/research/2026-05-21-maji-shadow-lesson-log-blob-slop.md:20 (resolved)
+
+**@copilot-pull-request-reviewer** (2026-05-21T01:52:09Z):
+
+P1: The sentence "Always use `git worktree`" reads as a universal recommendation, but `docs/HARNESS-SURFACES.md` notes the factory uses worktree isolation sparingly. Please soften/re-scope this guidance (e.g., recommend worktrees for multi-domain/long-running agent work or shared-environment risk) so it doesn’t contradict the current documented adoption stance.
+
+## General comments
+
+### @AceHack (2026-05-21T01:56:52Z)
+
+Vera coordination note: I rechecked this PR during the desktop loop tick. Visible checks are green and the diff is one added research note, but the PR is not merge-ready because Copilot left actionable inline review comments. Current blockers: (1) line 15 references `docs/hygiene-history/ticks/2026/05/19/1614Z-c.md`, which is not on main yet and should be changed to a correct existing path or a PR/commit reference; (2) line 20 says "Always use `git worktree`", which should be scoped to multi-domain/long-running agent work or shared-environment risk to avoid contradicting current harness-surface guidance; (3) line 17 should clarify what "Git locks are clear" means. Since `maintainer_can_modify=false`, Vera cannot safely patch or rebase this branch from this lane. Root checkout stayed read-only.
diff --git a/...ussions/PR-4482-rule-backlog-item-start-gate-add-pr-4478-as-full-file-verifi.md b/...ussions/PR-4482-rule-backlog-item-start-gate-add-pr-4478-as-full-file-verifi.md
@@ -0,0 +1,64 @@
+---
+pr_number: 4482
+title: "rule(backlog-item-start-gate): add PR #4478 as full-file verification empirical anchor"
+author: "AceHack"
+state: "MERGED"
+created_at: "2026-05-21T02:21:01Z"
+merged_at: "2026-05-21T02:22:14Z"
+closed_at: "2026-05-21T02:22:14Z"
+head_ref: "rule/cite-pr4478-as-full-file-verification-empirical-anchor-otto-cli-2026-05-21"
+base_ref: "main"
+archived_at: "2026-05-21T02:43:38Z"
+archive_tool: "tools/pr-preservation/archive-pr.ts"
+---
+
+# PR #4482: rule(backlog-item-start-gate): add PR #4478 as full-file verification empirical anchor
+
+## PR description
+
+## Summary
+
+[PR #4477](https://github.com/Lucent-Financial-Group/Zeta/pull/4477) landed the orphaned-branch discriminator section in [`backlog-item-start-gate.md`](.claude/rules/backlog-item-start-gate.md) with 3 empirical anchors. [PR #4478](https://github.com/Lucent-Financial-Group/Zeta/pull/4478) merged ~1 min after #4477 was authored — too late to be included in the original empirical-anchor list.
+
+This adds [PR #4478](https://github.com/Lucent-Financial-Group/Zeta/pull/4478) between [PR #4472](https://github.com/Lucent-Financial-Group/Zeta/pull/4472) and [PR #4205](https://github.com/Lucent-Financial-Group/Zeta/pull/4205) in chronological substrate-evolution order:
+
+1. [#4461](https://github.com/Lucent-Financial-Group/Zeta/pull/4461) — 0059Z cold-boot inheriting orphaned branch
+2. [#4472](https://github.com/Lucent-Financial-Group/Zeta/pull/4472) — 0149Z discriminator discovery (first-file spot-check)
+3. **[#4478](https://github.com/Lucent-Financial-Group/Zeta/pull/4478) — 0202Z full-file verification (this addition)**
+4. [#4205](https://github.com/Lucent-Financial-Group/Zeta/pull/4205) — peer-agent rescue the discriminator catches
+
+## Why this matters
+
+The full-file diff across all 32+4+4+4+2 files of the 5 orphaned commits confirmed the cheap first-file heuristic correctly classified 4 of 4 cases this session checked exhaustively. That's the **operational-reliability anchor** for the discriminator — the spot-check version is cheap and works, demonstrated on a real branch.
+
+## Substrate-honest framing
+
+Aaron + I converged independently on the same generalization within ~12 minutes (Aaron's rule landing + my full-file verification shard merging). The substrate-coordination was good. This citation closes the small chronological gap his PR had no way to know about at authoring time.
+
+## Test plan
+
+- [x] `git ls-tree HEAD` = 53 (canary check)
+- [x] Diff is 6 lines added (one paragraph insertion)
+- [x] No content changes to the discriminator itself — additive citation only
+- [ ] CI runs (docs-only)
+
+🤖 Generated with [Claude Code](https://claude.com/claude-code)
+
+## Reviews
+
+### COMMENTED — @copilot-pull-request-reviewer (2026-05-21T02:22:57Z)
+
+## Pull request overview
+
+Adds PR #4478 as an additional empirical anchor in the orphaned-branch triage discriminator documentation within `.claude/rules/backlog-item-start-gate.md`, strengthening the evidence trail for the “full-file verification” outcome.
+
+**Changes:**
+- Add PR #4478 link and explanatory parenthetical to the “Empirical anchors” list.
+
+## Review threads
+
+### Thread 1: .claude/rules/backlog-item-start-gate.md:125 (unresolved)
+
+**@copilot-pull-request-reviewer** (2026-05-21T02:22:57Z):
+
+This file now lists PR #4478 under "Empirical anchors", but the later "Composes with" bullet still hard-codes the old empirical-anchor trio (PRs #4205/#4461/#4472). Please update that later reference to include #4478 as well so the anchor list stays consistent within the rule.