backlog(B-0141): brittle-pointer auto-rewriter — section-number to anchor-link / stable-ID (Aaron 2026-05-01; ID reserved 2 days ago, never filed)

docs/BACKLOG.md

@@ -121,6 +121,7 @@ are closed (status: closed in frontmatter)._
 - [ ] **[B-0132](backlog/P2/B-0132-crdt-composition-for-bft-propagation-aaron-2026-05-01.md)** CRDT-composition for BFT propagation — substrate events as composed CRDTs
 - [ ] **[B-0133](backlog/P2/B-0133-sequent-calculus-for-claim-retraction-attribution-aaron-2026-05-01.md)** Sequent calculus / labeled deductive systems for claim/retraction/attribution
 - [ ] **[B-0134](backlog/P2/B-0134-type-theoretic-orthogonality-discipline-encoding-aaron-2026-05-01.md)** Type-theoretic encoding of orthogonality discipline (extension vs creation as decidable judgment)
+- [ ] **[B-0141](backlog/P2/B-0141-brittle-pointer-anchor-link-auto-rewriter-aaron-2026-05-01.md)** Brittle-pointer auto-rewriter — convert section-number pointers to anchor-links so markdown cross-references survive renumbering and partial-rename (Aaron 2026-05-01)
 - [ ] **[B-0143](backlog/P2/B-0143-bidirectional-messaging-aaron-text-imessage-whatsapp-2026-05-01.md)** Bi-directional messaging integration with Aaron — text/iMessage/WhatsApp/etc., cheapest-or-free first, budget-controlled (Aaron 2026-05-01)
 - [ ] **[B-0146](backlog/P2/B-0146-formal-architecture-ladder-explicit-layer-declaration-aaron-2026-05-01.md)** Formal architecture ladder — explicit-layer-declaration discipline for B-rows
 - [ ] **[B-0147](backlog/P2/B-0147-timeseries-db-native-in-zsets-multi-dsl-integration-research-aaron-2026-05-01.md)** TimeSeries DB native-in-Zsets multi-DSL integration research (metrics-are-our-eyes)

docs/backlog/P2/B-0141-brittle-pointer-anchor-link-auto-rewriter-aaron-2026-05-01.md

@@ -0,0 +1,98 @@
+---
+id: B-0141
+priority: P2
+status: open
+title: Brittle-pointer auto-rewriter — convert section-number pointers to anchor-links so markdown cross-references survive renumbering and partial-rename (Aaron 2026-05-01)
+tier: tooling
+effort: M
+ask: Aaron 2026-05-01 named the row in the parallelism-scaling-ladder memo (`feedback_parallelism_scaling_ladder_kenji_unlocked_loop_agent_doc_code_two_lane_file_isolation_peer_mode_claims_automated_best_practice_at_scale_aaron_2026_05_01.md`) — "B-0141, not yet filed" — as the guardrail-vs-mover row's mechanized response to brittle pointers. ID was reserved; per-row file was never filed (caught by Aaron 2026-05-03 reminder during B-0176 casing-fix discussion).
+created: 2026-05-01
+last_updated: 2026-05-03
+depends_on: []
+composes_with: [B-0170, B-0175, B-0177]
+tags: [pointer, anchor-link, brittle, mechanization, markdown, rename, cross-reference, auto-rewriter, tooling]
+---
+
+# Brittle-pointer auto-rewriter — section-number pointers to anchor-links
+
+## Origin
+
+Aaron 2026-05-01, in the parallelism-scaling-ladder memo (autonomous-loop maintainer channel), named this row in the mechanized-guardrail-vs-mover table:
+
+> | Brittle-pointer (B-0141, not yet filed) | Pre/post check fails | Auto-rewriter converts §N → anchor-link |
+
+Aaron 2026-05-03, during B-0176 title-casing review-cycle, surfaced the row was never filed:
+
+> "B-0176 lowercase-title vs B-0170 there is a baclog item for md links that survive rename where possible"
+>
+> "it might have the word technical numbering or somthing in it"
+
+Otto searched extensively across `docs/backlog/` and `memory/`, could not locate the row by direct file. Subsequent search via the parallelism-scaling-ladder memo's table-row revealed: the ID was reserved, the per-row file was never filed. Two days of substrate-time without the row materializing — exactly the failure mode B-0177 (audit memos for misfiled backlog) addresses.
+
+## The problem
+
+Markdown cross-references in this repo currently use multiple pointer styles:
+
+1. **Section-number references** — *"per §3.2"* / *"per CLAUDE.md §54"* / *"per GOVERNANCE.md §33"*
+2. **Heading-text anchor links** — `[text](path#heading-slug)` (slug from heading text)
+3. **Direct file paths** — `path/to/file.md` (no anchor; relies on file existence)
+4. **B-NNNN backlog IDs** — stable IDs that survive filename rename
+5. **BP-NN best-practice IDs** — stable rule IDs in `docs/AGENT-BEST-PRACTICES.md`
+
+**The brittleness ranking** (least brittle to most):
+
+| Form | Survives section-renumber | Survives heading-text-edit | Survives filename-rename |
+|---|---|---|---|
+| B-NNNN ID | Yes | N/A | Yes (ID is stable) |
+| BP-NN ID | Yes | N/A | Yes (ID is stable) |
+| Anchor-link `path#stable-slug` | Yes | No (slug derives from text) | No (path is file) |
+| Section-number `§3.2` | No (renumber breaks it) | Yes | Depends |
+| Direct file path | N/A | N/A | No |
+
+**Brittle-pointer** = form that breaks under one of these renames.
+
+## What this row builds
+
+An auto-rewriter that converts brittle section-number pointers to less-brittle forms where possible:
+
+1. **Detection**: regex / parser identifies `§N` / `§N.M` / `section N` patterns referencing other docs
+2. **Resolution**: for each detected pointer, look up the current heading at that section number
+3. **Rewrite**: replace `§N` with `[heading text](path#heading-slug)` OR with the more stable B-NNNN / BP-NN ID if applicable
+4. **Bidirectional check**: confirm the target anchor / ID exists post-rewrite
+
+Optional second pass:
+
+- For anchor-links pointing at headings, check heading-text stability (compare git-blame age + recent edit frequency)
+- For very-stable headings, leave anchor as-is; for unstable headings, suggest a stable-ID-based alternative
+
+## Composes with
+
+- **B-0170 (substrate-claim-checker)**: checks claims for drift; this row's tool checks pointers for brittleness — sibling-instance of mechanized-quality-tooling
+- **B-0175 (substrate-retrieval-index)**: layer-4 active retrieval; pointer-rewriting is a layer-3 mechanization that helps in-flight retrieval find correct targets
+- **B-0177 (audit memos for misfiled backlog rows)**: this row's existence IS empirical evidence that B-0177's hypothesis is correct (memo named `B-0141, not yet filed` — never got filed; B-0177 catches this class)
+- **memory/parallelism-scaling-ladder memo (Aaron 2026-05-01)**: the originating substrate where the ID was reserved
+
+## Why this is M-effort
+
+- Detection regex is small (~50 LOC)
+- Resolution requires markdown-AST parsing to find target headings (use existing markdown lib)
+- Rewrite logic non-trivial: heading-slug generation must match GitHub's algorithm; bidirectional check needs file-system + content access
+- Test fixtures need authoring
+- Composition with existing markdownlint passes; should run as a one-shot pass (NOT every commit, since rewrite is destructive)
+
+## Open design questions (NOT for this row; for the design pass)
+
+1. **Mechanized vs advisory**: should the auto-rewriter rewrite in place, OR emit a list of suggested rewrites for human/agent review?
+2. **Default form preference**: when both anchor-link and B-NNNN/BP-NN exist, which is preferred? (Stable-ID seems strictly better)
+3. **Cadence**: one-shot conversion + ongoing lint, OR scheduled-cadence repair pass?
+4. **Section-number residual cases**: some pointers genuinely want section-numbers (e.g., GOVERNANCE.md §N where the document is structured as a numbered list). When NOT to rewrite?
+
+## Why this matters
+
+Brittle pointers create review-cycle noise: a renumbered section breaks every cross-reference, requiring manual sweep. Stable-ID-based references (B-NNNN, BP-NN, anchor-links) survive renumbering. The auto-rewriter mechanizes the conversion that's currently done manually (and inconsistently).
+
+Composes with the alignment-frontier: substrate-quality tooling that prevents pointer-rot is alignment-frontier substrate. The pointers ARE part of the substrate's connective tissue; rotting tissue degrades the substrate's coherence over time.
+
+## Carved sentence
+
+**"Brittle-pointer rewriter converts §N section-number references to stable forms (B-NNNN ID, BP-NN ID, or anchor-link with heading-text slug) where possible. Stable-ID references (B-NNNN, BP-NN) survive both renumbering AND filename rename; anchor-links survive renumbering; section-numbers survive neither. The mechanized rewriter prevents pointer-rot from accumulating in the substrate's connective tissue."**