lozit · Oct 9, 2026 · Oct 9, 2026
diff --git a/.claude-plugin/marketplace.json b/.claude-plugin/marketplace.json
@@ -6,14 +6,14 @@
   },
   "metadata": {
     "description": "Mono-plugin marketplace for the groundrules Claude Code plugin (formerly starter-kit)",
-    "version": "1.12.0"
+    "version": "1.13.0"
   },
   "plugins": [
     {
       "name": "groundrules",
       "source": "./",
       "description": "Interactive bootstrap of Claude Code projects — docs structure + best practices of documentation and configuration",
-      "version": "1.12.0",
+      "version": "1.13.0",
       "category": "scaffold"
     }
   ]

diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
 {
   "name": "groundrules",
-  "version": "1.12.0",
+  "version": "1.13.0",
   "description": "Interactive bootstrap of Claude Code projects: interview, docs/ structure, best practices of documentation and configuration, git init, optional remote.",
   "author": {
     "name": "Guillaume Ferrari",

diff --git a/.claude/rules/plugin-meta.md b/.claude/rules/plugin-meta.md
@@ -4,7 +4,7 @@ paths:
   - "skills/**"
   - "**/*.tpl"
 ---
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 
 # Plugin meta-development rules
 

diff --git a/.github/workflows/adr-index.yml b/.github/workflows/adr-index.yml
@@ -1,4 +1,4 @@
-# generated-by: groundrules v1.12.0
+# generated-by: groundrules v1.13.0
 #
 # The ADR index invariant, checked where the omission lands: a pull request.
 #

diff --git a/.groundrules.json b/.groundrules.json
@@ -1,5 +1,5 @@
 {
-  "groundrulesVersion": "1.12.0",
+  "groundrulesVersion": "1.13.0",
   "bootstrappedAt": "2026-05-11",
   "bootstrappedWithVersion": "0.1.0",
   "migrations": [
@@ -164,6 +164,12 @@
       "to": "1.12.0",
       "at": "2026-09-11",
       "note": "V1.12 the loop's verifier becomes genuinely independent (ADR 0044): run-loop.sh makes TWO invocations per iteration — a maker pass that leaves its work uncommitted, then a separate verifier pass with a fresh context that judges the working tree and holds the write; validate-runner.sh asserts the split by counting invocations. The verifier preamble's 'nothing else' now bars the maker's narration rather than the artifacts its own checks require. AGENT-EVALS retires 'validated' for 'watching' / 'probed: <case>, N/N since <date>' (ADR 0043): a case covers one instance, never the behavioural class. The eval suite moves to skill-creator's harness, the CLI runner being gated (ADR 0042). The meta CLAUDE.md gains the Session start reading list the plugin ships to every project but never applied to itself — guards in LEARNINGS and AGENT-EVALS were unreadable here, hence unable to fire. No user-facing template change beyond the loop."
+    },
+    {
+      "from": "1.12.0",
+      "to": "1.13.0",
+      "at": "2026-10-09",
+      "note": "V1.13 optional INBOX.md at the repository root (ADR 0048) — what was handed to the project from outside, dated and verbatim, not yet triaged; offered by bootstrap and adopt, unchecked by default, and named FIRST in the generated CLAUDE.md session-start list (no conditional: the list already marks optional entries '(if present)'). The plugin never fills it — how lines arrive is the caller's business. The request behind it asked for a SessionStart hook reading one estate's private document layout and was refused on ADR 0025 (nothing the plugin generates runs unasked) and on the rule that it never describes one estate's layout; the principle recorded is the convention here, the transport elsewhere. AGENTS.md investigated and deferred to M2 with the harness behaviour measured (ADR 0047). Dogfood: the orchestrated working regime recorded (ADR 0046, amends 0036's commit rule), 'validated' retired for 'probed' (ADR 0043), the ADR index invariant made a CI check. No other change to generated output."
     }
   ],
   "answers": {

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,12 +1,12 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Changelog
 
 All notable changes to this project are documented in this file.
 
 Format inspired by [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 versions follow [Semantic Versioning](https://semver.org/).
 
-## [Unreleased]
+## [1.13.0] - 2026-10-09
 
 ### Added
 - **An optional `INBOX.md`: what was handed to this project from outside, not yet triaged** ([ADR 0048](docs/decisions/0048-the-convention-here-the-transport-elsewhere.md)). Offered by `bootstrap` and `adopt`, unchecked by default, at the repository root: one dated line per subject, **verbatim**, and triage **marks** a line (`→ ADR 0012`, `→ PLAN.md`, `→ dropped`) rather than deleting it — the file is a record of what arrived, not a queue that empties. It is named **first** in the generated `CLAUDE.md`'s session-start list, before `PLAN.md`, because what arrived can change what the rest of the list means; no conditional was needed, the list already marks optional entries *(if present)*. **The plugin never fills it** — how lines get there is the caller's business, described nowhere here.

diff --git a/CLAUDE.md b/CLAUDE.md
@@ -82,7 +82,7 @@ Reloading after a change: a **new skill directory** requires a full restart of C
 
 ## Template conventions
 
-- Every template carries `<!-- generated-by: groundrules v1.12.0 -->` at the top.
+- Every template carries `<!-- generated-by: groundrules v1.13.0 -->` at the top.
 - Placeholders in `{{KEY}}` format (plain text substitution, no template engine).
 - Available placeholders defined in `skills/bootstrap/SKILL.md` (Phase 5).
 - **English-only**: one `.tpl` per file, no language suffix (cf. ADR 0012). `gitignore.minimal` and `adr-template.md` are plain files.

diff --git a/PLAN.md b/PLAN.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # PLAN — groundrules
 
 **Active** plan/todo for the project. Maintained by Claude during work.

diff --git a/README.md b/README.md
@@ -216,6 +216,7 @@ Not everything here is a research result — and we won't pretend it is. Some ch
 | **Semantic Versioning** — `MAJOR.MINOR.PATCH` | A version number that *means something*: consumers can tell at a glance whether an update is a fix, a feature, or a breaking change. | [semver.org](https://semver.org/) |
 | **Test-Driven Development** — a failing test *before* the code (**loop regime only**) | When a task is handed to the autonomous loop (`/groundrules:realize`), a pre-written, currently-**red** acceptance test is the loop's back pressure: the maker makes it green, an independent verifier replays it. Red-first is Kent Beck's discipline; groundrules applies it as the *loop precondition* — **not** a global mandate (interactive work keeps the human as back pressure, handoff-not-gospel). | [Beck, *TDD by Example*](https://www.oreilly.com/library/view/test-driven-development/0321146530/) · [ADR 0027](docs/decisions/0027-reflection-realization-interactive-loop.md) |
 | **Reconciliation over reminders** — a doc you can't generate gets *checked* against the code | A rule that asks an agent to remember to update a record relies on attention and fails quietly; comparing the record with the code does not. Martraire calls this a **reconciliation mechanism**: where documentation can't be generated from the source, verify it against the source and report the mismatch. `/groundrules:close` applies it to `PLAN.md` — git says what changed, the plan says what it believes — and leaves the edit to a human. | [Martraire, *Living Documentation*](https://www.oreilly.com/library/view/living-documentation-continuous/9780134689418/) · [ADR 0038](docs/decisions/0038-close-reconcile-plan-against-diff.md) |
+| **Capture is free, deciding is not** — an inbox separate from the plan | Writing down what arrived and deciding what it means are different acts, and fusing them makes you refuse input to avoid the decision. Allen's capture/clarify split is the standard form; groundrules applies it as an optional `INBOX.md` holding dated, verbatim lines, named first in the generated `CLAUDE.md`'s session-start list — and triage **marks** a line rather than deleting it, so the file stays a record of what arrived rather than a queue that empties. | [Allen, *Getting Things Done*](https://gettingthingsdone.com/) · [ADR 0048](docs/decisions/0048-the-convention-here-the-transport-elsewhere.md) |
 | **Small, surgical diffs** — simplicity first, touch only what the task needs | The smallest change that solves the *stated* problem is easier to review, revert, and reason about; speculative abstraction and drive-by refactors are liabilities, not value (the YAGNI / minimal-change craft consensus). groundrules folds it into the generated `CLAUDE.md` **Posture**. | [YAGNI](https://martinfowler.com/bliki/Yagni.html) · [ADR 0034](docs/decisions/0034-posture-keep-the-diff-small.md) |
 
 These aren't dressed up as science — they're conventions we chose deliberately, and the choice itself is recorded where every other one is: [`docs/decisions/`](docs/decisions/).
@@ -266,6 +267,7 @@ The findings behind the design choices above:
 - **Semantic Versioning** — [semver.org](https://semver.org/)
 - **Test-Driven Development** — Kent Beck, *Test-Driven Development: By Example* (Addison-Wesley, 2002) — red-first as the loop's back pressure. [O'Reilly](https://www.oreilly.com/library/view/test-driven-development/0321146530/)
 - **Reconciliation mechanisms** — Cyrille Martraire, *Living Documentation: Continuous Knowledge Sharing by Design* (Addison-Wesley, 2019) — when a document can't be generated from the code, check it against the code and report the mismatch. [O'Reilly](https://www.oreilly.com/library/view/living-documentation-continuous/9780134689418/)
+- **Capture before clarifying** — David Allen, *Getting Things Done* (1st ed. 2001) — collect what arrives without judging it, decide what it means as a separate step. [gettingthingsdone.com](https://gettingthingsdone.com/)
 
 The complete per-decision history lives in [`docs/decisions/`](docs/decisions/) and [`CHANGELOG.md`](CHANGELOG.md).
 

diff --git a/docs/AGENT-EVALS.md b/docs/AGENT-EVALS.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Agent evals — groundrules
 
 > A log of the **agent's own** observed failure modes while developing this plugin —

diff --git a/docs/ARCHITECTURE.md b/docs/ARCHITECTURE.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Architecture — groundrules
 
 **Living** snapshot of the current architecture. Updated as the structure evolves.

diff --git a/docs/CONTEXT-ECONOMY.md b/docs/CONTEXT-ECONOMY.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Context economy — how groundrules keeps docs exhaustive *and* cheap
 
 > Why groundrules generates an **index + on-demand docs** instead of dumping everything

diff --git a/docs/LEARNINGS.md b/docs/LEARNINGS.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Learnings — groundrules
 
 Non-trivial learnings that emerged during the project. Reverse-chronological order (newest at the top).

diff --git a/docs/ROADMAP.md b/docs/ROADMAP.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # ROADMAP — groundrules
 
 Long-term direction: the **big milestones** we want to reach. This is the *vision* horizon —

diff --git a/docs/VISION.md b/docs/VISION.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Vision — groundrules
 
 > Synthesis of the project intent. Source: `intake/INTENT.md` (retroactive paste of the original prompt). Update when the intent evolves (rare; tactical decisions go in `docs/decisions/`).

diff --git a/docs/best-practices-pending.md b/docs/best-practices-pending.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # Best practices pending — to review manually
 
 File generated by `/starter-kit:apply-best-practices` on 2026-05-11. Contains the recommendations **not applied automatically** (hooks, custom skills) awaiting human review.

diff --git a/docs/decisions/0000-template.md b/docs/decisions/0000-template.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # NNNN — Short decision title
 
 **Date**: YYYY-MM-DD

diff --git a/docs/decisions/0001-marketplace-json-location.md b/docs/decisions/0001-marketplace-json-location.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0001 — `marketplace.json` lives in `.claude-plugin/`, not at the repo root
 
 **Date**: 2026-05-11

diff --git a/docs/decisions/0002-plain-text-placeholder-substitution.md b/docs/decisions/0002-plain-text-placeholder-substitution.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0002 — Plain text placeholder substitution, no templating engine
 
 **Date**: 2026-05-11

diff --git a/docs/decisions/0003-multi-skill-architecture.md b/docs/decisions/0003-multi-skill-architecture.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0003 — Multi-skill architecture with `disable-model-invocation: true`
 
 **Date**: 2026-05-11

diff --git a/docs/decisions/0004-starter-kit-json-schema.md b/docs/decisions/0004-starter-kit-json-schema.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0004 — `.starter-kit.json` schema with `bootstrappedWithVersion` and `migrations` array
 
 **Date**: 2026-05-11

diff --git a/docs/decisions/0005-intent-capture-and-apply-best-practices.md b/docs/decisions/0005-intent-capture-and-apply-best-practices.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0005 — Intent capture in bootstrap + separate `apply-best-practices` skill
 
 **Date**: 2026-05-11

diff --git a/docs/decisions/0006-optional-specialized-docs.md b/docs/decisions/0006-optional-specialized-docs.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0006 — Optional specialized docs in bootstrap
 
 **Date**: 2026-06-03

diff --git a/docs/decisions/0007-denumber-entry-docs.md b/docs/decisions/0007-denumber-entry-docs.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0007 — Drop the numeric prefix on entry docs (VISION / INTENT)
 
 **Date**: 2026-06-03

diff --git a/docs/decisions/0008-adopt-brownfield-projects.md b/docs/decisions/0008-adopt-brownfield-projects.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0008 — Adopt existing (brownfield) projects via a dedicated skill
 
 **Date**: 2026-06-03

diff --git a/docs/decisions/0009-global-claude-md-awareness.md b/docs/decisions/0009-global-claude-md-awareness.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0009 — Global / enterprise CLAUDE.md awareness + lean project CLAUDE.md
 
 **Date**: 2026-06-03

diff --git a/docs/decisions/0010-managed-project-claude-md-deference.md b/docs/decisions/0010-managed-project-claude-md-deference.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0010 — Defer to an existing tool-managed project CLAUDE.md
 
 **Date**: 2026-06-03

diff --git a/docs/decisions/0011-detect-no-ai-attribution-policy.md b/docs/decisions/0011-detect-no-ai-attribution-policy.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0011 — Detect a "no AI attribution" policy and adapt suggested commits
 
 **Date**: 2026-06-03

diff --git a/docs/decisions/0012-english-only.md b/docs/decisions/0012-english-only.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0012 — English-only (drop bilingual FR/EN templates)
 
 **Date**: 2026-06-04

diff --git a/docs/decisions/0013-media-under-docs.md b/docs/decisions/0013-media-under-docs.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0013 — Move media under docs/ (docs/media/ instead of top-level media/)
 
 **Date**: 2026-06-04

diff --git a/docs/decisions/0014-rename-brief-to-intake.md b/docs/decisions/0014-rename-brief-to-intake.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0014 — Rename the brief/ folder to intake/
 
 **Date**: 2026-06-06

diff --git a/docs/decisions/0015-best-effort-update-check.md b/docs/decisions/0015-best-effort-update-check.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0015 — Best-effort plugin update check in skills (Phase 0)
 
 **Date**: 2026-06-06

diff --git a/docs/decisions/0016-rename-marketplace.md b/docs/decisions/0016-rename-marketplace.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0016 — Rename the marketplace starter-kit-local → claude-code-starter-kit
 
 **Date**: 2026-06-06

diff --git a/docs/decisions/0017-plugin-rename-at-v1.md b/docs/decisions/0017-plugin-rename-at-v1.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0017 — Rename the plugin to groundrules at V1.0.0
 
 **Date**: 2026-06-06

diff --git a/docs/decisions/0018-adopt-consolidation-mode.md b/docs/decisions/0018-adopt-consolidation-mode.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0018 — adopt: map-in-place vs consolidate strategies
 
 **Date**: 2026-06-06

diff --git a/docs/decisions/0019-heyjoe-inspired-doc-improvements.md b/docs/decisions/0019-heyjoe-inspired-doc-improvements.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0019 — Doc improvements harvested from a real project (crm-heyjoe)
 
 **Date**: 2026-06-06

diff --git a/docs/decisions/0020-repo-is-the-only-memory.md b/docs/decisions/0020-repo-is-the-only-memory.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0020 — The repo is the only memory
 
 **Date**: 2026-06-07

diff --git a/docs/decisions/0021-context-economy-index-over-doc-search.md b/docs/decisions/0021-context-economy-index-over-doc-search.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0021 — Context economy: index over doc-search for a project's own docs
 
 **Date**: 2026-06-08

diff --git a/docs/decisions/0022-agent-evals-and-session-close.md b/docs/decisions/0022-agent-evals-and-session-close.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0022 — Session-close ritual + optional agent-evals log
 
 **Date**: 2026-06-08

diff --git a/docs/decisions/0023-project-scope-for-team-portability.md b/docs/decisions/0023-project-scope-for-team-portability.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0023 — Recommend project-scope install for team portability (no CLAUDE.md bloat, no vendoring)
 
 **Date**: 2026-06-08

diff --git a/docs/decisions/0024-slim-skill-claude-md-budget.md b/docs/decisions/0024-slim-skill-claude-md-budget.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0024 — `/groundrules:slim` — operationalize the CLAUDE.md budget
 
 **Date**: 2026-06-08

diff --git a/docs/decisions/0025-no-runtime-hook-no-watch.md b/docs/decisions/0025-no-runtime-hook-no-watch.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0025 — Won't-do: PreToolUse `{{KEY}}` hook and `/watch-bootstrap`
 
 **Date**: 2026-06-08

diff --git a/docs/decisions/0026-posture-and-per-feature-prd.md b/docs/decisions/0026-posture-and-per-feature-prd.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0026 — Posture section + per-feature PRD (harvested from the four-principles doc)
 
 **Date**: 2026-06-13

diff --git a/docs/decisions/0027-reflection-realization-interactive-loop.md b/docs/decisions/0027-reflection-realization-interactive-loop.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0027 — Reflection vs realization phases; interactive vs loop regimes
 
 **Date**: 2026-06-13

diff --git a/docs/decisions/0028-git-workflow-conventions.md b/docs/decisions/0028-git-workflow-conventions.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0028 — Git workflow conventions (branching, commit granularity, AI attribution)
 
 **Date**: 2026-06-13

diff --git a/docs/decisions/0029-content-aware-claude-md-tailoring.md b/docs/decisions/0029-content-aware-claude-md-tailoring.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0029 — Content-aware CLAUDE.md tailoring (retire the lean template)
 
 **Date**: 2026-06-14

diff --git a/docs/decisions/0030-loop-namespace-and-backlog.md b/docs/decisions/0030-loop-namespace-and-backlog.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0030 — `loop/` as the loop's namespace; a dedicated backlog referenced by `PLAN.md`
 
 **Date**: 2026-06-14

diff --git a/docs/decisions/0031-goal-interop-swappable-loop-executor.md b/docs/decisions/0031-goal-interop-swappable-loop-executor.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0031 — `/goal` interop: the loop executor is swappable; reposition `run-loop.sh` as fallback
 
 **Date**: 2026-06-15

diff --git a/docs/decisions/0032-premortem-anti-sycophancy.md b/docs/decisions/0032-premortem-anti-sycophancy.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0032 — Adopt the premortem as a reflection-phase adversarial technique (anti-sycophancy)
 
 **Date**: 2026-06-20

diff --git a/docs/decisions/0033-adopt-full-mode.md b/docs/decisions/0033-adopt-full-mode.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0033 — adopt: "Full adoption" mode (consolidate + reformat, single grouped confirmation)
 
 **Date**: 2026-06-26

diff --git a/docs/decisions/0034-posture-keep-the-diff-small.md b/docs/decisions/0034-posture-keep-the-diff-small.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0034 — Posture: a "keep the diff small" axis (simplicity + surgical changes)
 
 **Date**: 2026-06-29

diff --git a/docs/decisions/0035-no-auto-capture-memory-layer.md b/docs/decisions/0035-no-auto-capture-memory-layer.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0035 — Considered an auto-capture memory layer (claude-mem class); declined
 
 **Date**: 2026-06-29

diff --git a/docs/decisions/0036-git-workflow-corrected.md b/docs/decisions/0036-git-workflow-corrected.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0036 — Git workflow corrected: PR on protected `main`, no AI attribution (supersedes 0028)
 
 **Date**: 2026-09-02

diff --git a/docs/decisions/0037-executable-evals-over-agent-config.md b/docs/decisions/0037-executable-evals-over-agent-config.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0037 — An executable eval suite over the agent's configuration (`evals/`), native format, out of band
 
 **Date**: 2026-09-02

diff --git a/docs/decisions/0038-close-reconcile-plan-against-diff.md b/docs/decisions/0038-close-reconcile-plan-against-diff.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0038 — `/groundrules:close`: reconcile `PLAN.md` against the diff — a check, not a reminder
 
 **Date**: 2026-09-08

diff --git a/docs/decisions/0039-conditional-superpowers-interop.md b/docs/decisions/0039-conditional-superpowers-interop.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0039 — The superpowers interop section is conditional, detected rather than asked
 
 **Date**: 2026-09-08

diff --git a/docs/decisions/0040-three-bootstrap-modes.md b/docs/decisions/0040-three-bootstrap-modes.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0040 — Three bootstrap modes: emptiness and prior state are two questions, not one
 
 **Date**: 2026-09-08

diff --git a/docs/decisions/0041-a-migration-never-damages-a-correct-file.md b/docs/decisions/0041-a-migration-never-damages-a-correct-file.md
@@ -1,4 +1,4 @@
-<!-- generated-by: groundrules v1.12.0 -->
+<!-- generated-by: groundrules v1.13.0 -->
 # 0041 — A migration never damages a correct file: two write guards, and a comparison text that matches what `bootstrap` would produce
 
 **Date**: 2026-09-08