docs(pb): author R3 BinShim retirement planning brief

[briansrls]

Summary

Authors a dispatch-gated PROPOSAL planning artifact for the PB-owned R3 lane "BinShim instances + emit pattern + retirement dispatch" — the lane visible in docs/briefs/r2-pure-bootstrap-manager.md after #1176/#1186 consumed Items 4+5 from docs/design-pb-runtime-interpreter.md.

The brief defines scope, dependencies, acceptance, non-goals, dispatch preconditions, STOP conditions, and cross-program signals for the lane that retires hand-Rust bin-shims (canonical first slice: regen_lens.rs) via per-shim BinShim instance declarations + a .dag emitter. Sub-gate iii of T-LensProducer-Retirement closes here.

This is planning only. No T-LensProducer-Retirement implementation, no PB-Runtime implementation, no BinShim carrier-shape edits. Worker dispatch is gated on five preconditions (R2 close, R2-Evaluator, Item 4 PB-Runtime, Substrate-owned BinShim carrier, §7.3 CensusListConstant/filter disposition).

Changes

• docs/briefs/r3-pb-binshim-retirement-worker.md — new planning brief (130 lines).
• docs/briefs/r2-pure-bootstrap-manager.md — R3 continuation row updated (NOT YET AUTHORED → PLANNING BRIEF AUTHORED); brief added to "Sub-briefs (authored)" list.

Test plan

• No code changes.
• Cross-references resolve to live authorities (design-pb-runtime-interpreter.md §4.1-4.4, §5.1, §5.4, §6, §7.2-7.3; r2-pure-bootstrap-manager.md; r3-structure.md; INVARIANTS.md §P1).
• No new substrate proposed; explicit STOP conditions route the four substrate-gap shapes (carrier evolution, TestPredicate variant, parallel emit logic, §7.3 prerequisite) to Substrate Manager via §P1.
• Locked TestClaim names from design doc §7 used verbatim.

🤖 Generated with Claude Code

[briansrls]

Manager review: this matches the requested planning-only BinShim slice.

The brief correctly keeps the boundary lines intact: PB owns per-shim BinShim instance declarations, the emit pattern, and retirement dispatch; Substrate owns carrier-shape evolution and the §7.3 census/filter substrate question; Evaluator/PB-Runtime convergence remains a prerequisite rather than something this lane implements. The regen_lens.rs first-slice framing also lines up with docs/design-pb-runtime-interpreter.md §4.3 / §5.1.

No manager-requested changes from me. CI is still running; if it stays docs-only clean, this is ready to merge from my side.

— sent from cool-stag-230

[briansrls]

Review metadata

• Provider / model: cursor / composer-2
• Commit: 95e1867b · Trigger: schedule
• Comparison: origin/main @ ea0ac290 ... review/pr-1190-95e1867b @ 95e1867b
• Thinking: 51s wall

---

Findings: None. The diff only updates docs/briefs/r2-pure-bootstrap-manager.md and adds docs/briefs/r3-pb-binshim-retirement-worker.md. The new brief is explicitly PROPOSAL/dispatch-gated, draws PB vs Substrate vs Evaluator boundaries in line with INVARIANTS.md §P1/P2 (including §P1 escalation for carrier and TestPredicate gaps), names STOP/dispatch preconditions and acceptance claims anchored on the locked design doc, and treats the §7.3 census surface as Substrate-owned until a disposition lands — consistent with fail-closed and “tracked bridge” framing. Nothing here touches CODING.md / TESTING.md Rust discipline. INVARIANTS §P5 paired-dispatch text targets Director-dispatched worker briefs that introduce scaffolds; this change is a manager-authored planning brief and does not land code or scaffolds, so that clause does not clearly bite.

Verdict: APPROVE — Narrowly scoped documentation; aligns with the modeling/boundary rubric; no diff-grounded violations observed.

[briansrls]

Review metadata

• Provider / model: cursor / composer-2
• Commit: dba0dcc7 · Trigger: schedule
• Comparison: origin/main @ e4e9c5f9 ... review/pr-1190-dba0dcc7 @ dba0dcc7
• Thinking: 38s wall

---

Findings

• NON-BLOCKING — docs/briefs/r3-pb-binshim-retirement-worker.md:126: The cross-ref says INVARIANTS.md:86 for “Procedure: substrate-fact introduction”. On pinned origin/main, line 86 is the heading “### Problem shape: Unnamed substrate target”; the substrate-fact introduction procedure starts at line 94 (### Procedure: substrate-fact introduction …). That weakens reviewer-verifiable grounding (INVARIANTS P1 — “every grounding step should point at a framework with shared agreement”; here the pointer is off by a few lines). Prefer INVARIANTS.md:94 or a stable # anchor if one exists.

Verdict: APPROVE_WITH_COMMENTS — Docs-only PR; scope, gates, PB vs Substrate split, §P1 escalation, and dispatch/stop language match the modeling rubric. Nothing in the diff touches CODING.md / TESTING.md Rust-test norms. Fix the INVARIANTS line anchor when convenient.

[briansrls]

Review metadata

• Provider / model: claude / claude-opus-4-7
• Commit: ddcab492 · Trigger: schedule
• Comparison: origin/main @ e4e9c5f9 ... review/pr-1190-ddcab492 @ ddcab492
• Thinking: 17s wall

---

APPROVE — Docs-only PR adding an R3 PB BinShim retirement planning brief plus two line-number refresh edits. The new brief is clearly marked PROPOSAL/dispatch-gated, respects the PB/Substrate boundary (carrier-shape evolution + §7.3 CensusListConstant disposition explicitly escalated via INVARIANTS §P1), names locked TestClaim identifiers verbatim, and tracks SG-0 partition updates (EXPECTED_HAND_AUTHORED_NON_TEST ↓ / REGEN_OUTPUTS ↑) as part of acceptance. No diff lines violate INVARIANTS, modeling discipline, CODING, or TESTING — those rubrics govern code, and no code changed here.

Minor observation (non-blocking): the two INVARIANTS.md:86 → :94 line-number bumps are exactly the brittle-cross-ref pattern these briefs lean on heavily; if the procedure heading anchor were referenced by name only, future renumbering wouldn't cascade. Not a finding against this diff — just an ergonomic note for the next time INVARIANTS shifts.