mandrel 2.56.0 → 2.57.0

package/.agents/workflows/mandrel-plan.md

@@ -56,11 +56,13 @@ node .agents/scripts/plan-context.js --seed "<seed>" \
 and derives source ids from its `sourceTickets[]`; it also writes
 **`stories.template.json`**, step 2's skeleton.
 
-The envelope carries docs context, the story-author prompt, `sourceTickets[]`,
+The envelope carries docs context, the story-author prompt (`systemPrompts.story`,
+plus `systemPrompts.storySplitRules` for an N>1 draft), `sourceTickets[]`,
 `duplicates[]` (open **Stories**, never Epics), `epicCandidates[]` +
 `dependencyCandidates[]` (Gate #3; path collisions), `priorFeedback` and
-advisory `complexitySignals` (**no routing authority**). A trivial scope claims
-the lite route at persist, failing closed to `full`.
+advisory `complexitySignals` (**no routing authority**). An envelope over the
+planner-context ceiling is written truncated with a `truncated` note, never
+refused.
 
 **Triage each unknown by resolver** ([ref](helpers/plan-reference.md)): an
 **AFK** unknown (research settles it) is resolved before authoring, never
@@ -68,16 +70,13 @@ assumed; a **HITL** unknown goes to Gate #1. Under `--yes` do not ask free-form
 operator questions — AFK unknowns are still researched; only HITL unknowns land
 in Key Assumptions, each a decision-made-by-default.
 
-**Gate #1** — STOP to confirm the sharpened plan intent, any
-duplicate-candidate review, and any `intake` row worth graduating
-([ref](helpers/plan-reference.md)). Under `--yes`, auto-proceed.
-
-On a truthy `deliverLightSuggestion.suggested`, offer — advisory, never an
-automatic reroute — to deliver the seed instead; on confirm route **in this
-session** into [`helpers/deliver-light.md`](helpers/deliver-light.md), its gate
-filled from this envelope. A truthy `complexitySignals.uiSurface` names
-[`/prototype`](prototype.md); never invoke it here.
-[Both](helpers/plan-reference.md).
+**Gate #1** — STOP for exactly two things: confirm the sharpened plan intent,
+and settle any HITL unknown the operator owns. Everything else the envelope
+surfaced — `duplicates[]`, open `intake` rows, a truthy
+`memoryPoolAdvisory.recommend`, a truthy `complexitySignals.uiSurface` naming
+[`/prototype`](prototype.md) (never invoke it here) — collapses to
+**one advisory line** under the gate; none of it stops the run or reroutes
+it ([ref](helpers/plan-reference.md)). Under `--yes`, auto-proceed.
 
 ### 2. Author
 
@@ -93,34 +92,19 @@ included. One rescue: a **never-tracked** one normalises to `creates`. Fields:
 [ref](helpers/plan-reference.md).
 
 Artifacts under `temp/plan-<slug>/`: `stories.json` (**length 1 by default**;
-over-budget Specs fail closed — split or tighten, never under `docs/`); optional
+a Spec is as long as the work needs, inline, never under `docs/`); optional
 `techspec.md` (**N===1 only**, folded into `## Spec`) and
 `acceptance-manifest.json` (N>1 — `--plan-acceptance`). Use the envelope
-`systemPrompts.story`; split only under the policy above.
+`systemPrompts.story`; split only under the policy above, and when you do,
+read `systemPrompts.storySplitRules` too — it carries the schedule and
+partition rules the core omits.
 
 **Tickets mode:** every Story authors a top-level `supersedes[]`; persist
 refuses a partial map ([shape](helpers/plan-reference.md)).
 
-### 2.5 Critics
-
-```bash
-node .agents/scripts/plan-critics.js \
-  --stories temp/plan-<slug>/stories.json \
-  [--tech-spec temp/plan-<slug>/techspec.md]
-```
-
-Run **before** persist — the last point a finding folds into a re-author. It
-exits 0 on **any** verdict (verdicts route work, they do not gate) and exits
-**1** only on a usage/IO error — no critic ran: **do not proceed to Persist**,
-fix and re-run.
-
-- **Both `dispatch: false`** — proceed to Persist (each skip is ledgered).
-- **Either `dispatch: true`** — dispatch **one fresh-context, maker-blind
-  sub-agent per firing critic** (hand it only the draft artifacts, never the
-  authoring transcript), fold findings into Gate #2 or a re-author round, re-run
-  this step. Pre-mortem triggers (incl. the external-dependency probe), the
-  advisory-only `textHygiene.findings[]` lints and dispatch shape:
-  [reference](helpers/plan-reference.md).
+The maker-blind **pre-mortem** critic is not a step of this spine: run
+`plan-critics.js` only when the operator asks for it
+([how](helpers/plan-reference.md)).
 
 ### 3. Persist
 
@@ -132,7 +116,11 @@ to review (`--force-review`). Under `--yes`, auto-proceed.
 `--epic-goal`). Never unasked ([ref](helpers/plan-reference.md)).
 
 Run persist `--dry-run` **first** — same command, writes suppressed; every gate
-runs before the first `createIssue` ([list](helpers/plan-reference.md)):
+runs before the first `createIssue`, and the run **lists its warnings**
+(a `creates` / `refactors-existing` the base branch disagrees with, a goal or
+acceptance path absent at base, an open question in a body) and the
+`changes[]` repairs it applied ([list](helpers/plan-reference.md)). Read
+them; they never stop the persist:
 
 ```bash
 node .agents/scripts/plan-persist.js \
@@ -144,8 +132,8 @@ node .agents/scripts/plan-persist.js \
   [--epic <id> | --epic-title "<name>" --epic-goal "<one paragraph>"]
 ```
 
-At lite shape `--chain-on-clean` folds a clean dry-run into the persist; a full
-plan keeps its review trip.
+`--chain-on-clean` folds a clean dry-run into the persist for **any** plan —
+the dry-run's warning list is the review.
 
 Persist creates `type::story` issue(s), a **metadata-only** `plan-run::<id>`
 label, `blocked by #<id>` footers for every `depends_on` edge, and on a Gate #3
@@ -155,8 +143,7 @@ also comments on and closes each source id ([ref](helpers/plan-reference.md)).
 
 ## Constraints
 
-- `/mandrel-plan` starts delivery **only** through a confirmed Gate #1 light
-  route — never off its Stories, which land via
+- `/mandrel-plan` never starts delivery — its Stories land via
   [`/mandrel-deliver`](mandrel-deliver.md).
 - Duplicate search targets open Stories (`type::story`), not Epics; and
   deterministic gates still fail closed under `--yes`.

package/.agents/workflows/memory-consolidate.md

@@ -97,17 +97,14 @@ Then write the receipt to `.consolidation-stamp.json` in the pool root:
 ```
 
 `entryCount` is the surviving non-index `*.md` count **after** the rewrite —
-count the directory, never the plan. It is the baseline the next run measures
-growth against, so a wrong number silently mis-arms the nudge.
-
-The `/mandrel-plan` Phase 0 advisory re-arms on exactly three conditions: the
-stamp aging past `planning.memoryPool.staleAfterDays` (30),
-`planning.memoryPool.growthDelta` (25) entries written since that count, or
-`MEMORY.md` exceeding `planning.memoryPool.indexByteCeiling` (24576) bytes.
-Pool size alone never triggers it — a pass that keeps every entry still quiets
-the first two arms. A stamp with no `entryCount` leaves growth unmeasured, and
-only the age and index arms can speak until the next pass writes one; a stamp
-dated in the future reads as no stamp at all.
+count the directory, never the plan. The stamp is the operator's record of
+the pass; nothing in the framework reads it back.
+
+The `/mandrel-plan` Phase 0 advisory re-arms on exactly one condition:
+`MEMORY.md` exceeding `planning.memoryPool.indexByteCeiling` (24576) bytes —
+the harness cap past which the index it loads is truncated. Pool size and
+stamp age never trigger it (Story #5312 retired those arms): a pass that
+rewrites long index lines short quiets it without pruning an entry.
 
 Write it **only** after Gate #2 — the stamp asserts an operator reviewed the
 pass, so writing it early makes it a lie.
@@ -116,8 +113,7 @@ Close with counts: entries read, corrected, merged, pruned, the new total, and
 **the rewritten `MEMORY.md`'s size in bytes beside that count** — the index is
 truncated at the byte ceiling, so a pass that pruned entries but left the
 index over the cap has not fixed the loss, and the number is the only way the
-operator can see that. Then the forecast the operator would otherwise derive
-by hand: when the advisory next fires, and which arm reaches it first.
+operator can see that.
 
 ## Constraints
 

package/docs/CHANGELOG.md

@@ -15,6 +15,29 @@ All notable changes to this project will be documented in this file.
 -->
 <!-- markdownlint-disable-file MD004 MD012 MD037 -->
 
+## [2.57.0](https://github.com/dsj1984/mandrel/compare/mandrel-v2.56.0...mandrel-v2.57.0) (2026-09-12)
+
+
+### ⚠ BREAKING CHANGES
+
+* the .agentrc keys delivery.routing.freshCriticSampleRate, delivery.codeReview.maxFixScopeFiles, delivery.signals and delivery.quality.codingGuardrails.cyclomaticMustFix are removed; a 2.57.0 migration step strips them on upgrade.
+* the .agentrc.json keys planning.complexityGate, planning.riskHeuristics, planning.failOnSharedEditors, planning.requireExplicitCrossStoryDeps, planning.failOnRegistryConflicts, planning.failOnLargeFanOut, planning.largeFanOutThreshold, planning.crossCuttingRegistries, planning.memoryPool.staleAfterDays and planning.memoryPool.growthDelta are removed; the planning block is strict, so a config still carrying one fails validation until the 2.57.0 migration step strips it on `mandrel update`. plan-persist.js no longer accepts --route-downgrade-reason, --allow-over-budget or --allow-large-fan-out, and a verify[] entry is a bare command with no tier suffix.
+
+### Added
+
+* delivery diet: drop the sampling, round-ceiling, fix-scope, signal and cyclomatic knobs, script the ceremony derivation, let a bare test run earn credit, and make the light gate and footprint guard read the diff ([#5313](https://github.com/dsj1984/mandrel/issues/5313)) ([#5319](https://github.com/dsj1984/mandrel/issues/5319)) ([33fa9db](https://github.com/dsj1984/mandrel/commit/33fa9db4d01d8932bd886d380fe32b29db9a7bce))
+* planning diet: delete the sizing, spec, lite-route and verify-tier limits, demote the footprint probes to warnings, and render the author prompt for the one-Story default ([#5312](https://github.com/dsj1984/mandrel/issues/5312)) ([#5318](https://github.com/dsj1984/mandrel/issues/5318)) ([cbcfdac](https://github.com/dsj1984/mandrel/commit/cbcfdacbca2cce9731e12c6cdfdab095d23b3e12))
+
+
+### Fixed
+
+* score the CRAP worker path through the escomplex AST shim, and report an unscorable file as unscorable ([#5311](https://github.com/dsj1984/mandrel/issues/5311)) ([#5314](https://github.com/dsj1984/mandrel/issues/5314)) ([0a8e8c8](https://github.com/dsj1984/mandrel/commit/0a8e8c8cd8afb7af88b3d0d5a49e6359a91f3524))
+
+
+### Changed
+
+* free the CLI entrypoints' logic from `main` so it can be tested, and tighten the CRAP floor back to 2 ([#5316](https://github.com/dsj1984/mandrel/issues/5316)) ([#5317](https://github.com/dsj1984/mandrel/issues/5317)) ([f67cfb3](https://github.com/dsj1984/mandrel/commit/f67cfb3b6e531c38300690e4dcd16a4de7dd18f8))
+
 ## [2.56.0](https://github.com/dsj1984/mandrel/compare/mandrel-v2.55.0...mandrel-v2.56.0) (2026-09-11)
 
 

package/lib/migrations/index.js

@@ -59,6 +59,8 @@ import { retireEpicAcTags } from './steps/2.2.0-retire-epic-ac-tags.js';
 import { retireMaxSeedWords } from './steps/2.11.0-retire-max-seed-words.js';
 import { retireCodebaseSnapshot } from './steps/2.20.0-retire-codebase-snapshot.js';
 import { retireLintBaselineCommand } from './steps/2.32.0-retire-lint-baseline-command.js';
+import { retireDeliveryLimitKnobs } from './steps/2.57.0-retire-delivery-limit-knobs.js';
+import { retirePlanningLimitKnobs } from './steps/2.57.0-retire-planning-limit-knobs.js';
 
 /**
  * Ordered registry of migration steps. MUST stay sorted ascending by
@@ -78,6 +80,8 @@ export const migrations = [
   retireMaxSeedWords,
   retireCodebaseSnapshot,
   retireLintBaselineCommand,
+  retirePlanningLimitKnobs,
+  retireDeliveryLimitKnobs,
 ];
 
 /**

package/lib/migrations/steps/2.57.0-retire-delivery-limit-knobs.js

@@ -0,0 +1,45 @@
+// lib/migrations/steps/2.57.0-retire-delivery-limit-knobs.js
+/**
+ * Story #5313 — the delivery diet. Strip the retired `delivery.*` keys from a
+ * consumer's config:
+ *
+ *   - `delivery.routing.freshCriticSampleRate` — the maker-checker sampling
+ *     floor is gone; the standard profile routes purely off the derived
+ *     change level (high or underivable → fresh critic, low → inline).
+ *   - `delivery.codeReview.maxFixScopeFiles` — the auto-fix file-count
+ *     ceiling bounded remediation by count rather than by risk.
+ *   - `delivery.signals` (`rework.editsPerFile`, `retry.repeatCount`) — the
+ *     detector thresholds and `SIGNALS_DEFAULTS` are retired wholesale.
+ *   - `delivery.quality.codingGuardrails.cyclomaticMustFix` — the cyclomatic
+ *     ratchet keeps its fixed ceiling of 12; `cyclomaticFlag` stays advisory.
+ *
+ * Every affected block carries `additionalProperties: false`, so a config
+ * still setting any of them fails validation on upgrade rather than warning.
+ * It sweeps **both** config surfaces (`.agentrc.json` and the gitignored
+ * `.agentrc.local.json`), because `config-resolver.js` deep-merges the
+ * overlay before the AJV gate runs.
+ *
+ * Pruning: `delivery.signals` is removed whole (`pruneDepth: 1` prunes an
+ * emptied `delivery`); the nested keys prune their emptied ancestors up to
+ * `delivery` itself. `delivery` is optional, so an emptied block is removed
+ * rather than left as `{}`. A sibling key that survives keeps its block.
+ */
+
+import { createRetireAgentrcKeyStep } from '../helpers/retire-agentrc-key.js';
+
+export const retireDeliveryLimitKnobs = createRetireAgentrcKeyStep({
+  version: '2.57.0',
+  description:
+    'strip the retired delivery.* limit knobs from .agentrc.json — ' +
+    'routing.freshCriticSampleRate, codeReview.maxFixScopeFiles, signals, ' +
+    'and quality.codingGuardrails.cyclomaticMustFix (Story #5313)',
+  keys: [
+    { path: ['delivery', 'routing', 'freshCriticSampleRate'], pruneDepth: 2 },
+    { path: ['delivery', 'codeReview', 'maxFixScopeFiles'], pruneDepth: 2 },
+    { path: ['delivery', 'signals'], pruneDepth: 1 },
+    {
+      path: ['delivery', 'quality', 'codingGuardrails', 'cyclomaticMustFix'],
+      pruneDepth: 3,
+    },
+  ],
+});

package/lib/migrations/steps/2.57.0-retire-planning-limit-knobs.js

@@ -0,0 +1,59 @@
+// lib/migrations/steps/2.57.0-retire-planning-limit-knobs.js
+/**
+ * Story #5312 — the planning diet. Strip the ten retired `planning.*` keys
+ * from a consumer's config:
+ *
+ *   - `planning.complexityGate` — the plan-side lite claim and its persist
+ *     backstop are gone; nothing reads the switch.
+ *   - `planning.riskHeuristics` — the phrase list was empty in every consumer
+ *     that resolved it, and the critic trigger that matched it went with the
+ *     consolidation critic.
+ *   - `planning.failOnSharedEditors`, `planning.requireExplicitCrossStoryDeps`,
+ *     `planning.failOnRegistryConflicts`, `planning.failOnLargeFanOut`,
+ *     `planning.largeFanOutThreshold`, `planning.crossCuttingRegistries` —
+ *     every conflict finding is advisory now, and the registry and fan-out
+ *     findings no longer exist.
+ *   - `planning.memoryPool.staleAfterDays`, `planning.memoryPool.growthDelta`
+ *     — the memory-hygiene advisory keeps only its index-byte arm.
+ *
+ * The `planning` block carries `additionalProperties: false`, so a config
+ * still setting any of them fails validation on upgrade rather than warning.
+ * It sweeps **both** config surfaces (`.agentrc.json` and the gitignored
+ * `.agentrc.local.json`), because `config-resolver.js` deep-merges the
+ * overlay before the AJV gate runs.
+ *
+ * Pruning: the two `memoryPool` keys prune the `memoryPool` object when it
+ * empties and then `planning` itself (`pruneDepth: 2`); the top-level keys
+ * prune `planning` (`pruneDepth: 1`). `planning` is optional, so an emptied
+ * block is removed rather than left as `{}`. A `memoryPool` that still
+ * carries `indexByteCeiling` survives untouched.
+ */
+
+import { createRetireAgentrcKeyStep } from '../helpers/retire-agentrc-key.js';
+
+const TOP_LEVEL_KEYS = [
+  'complexityGate',
+  'riskHeuristics',
+  'failOnSharedEditors',
+  'requireExplicitCrossStoryDeps',
+  'failOnRegistryConflicts',
+  'failOnLargeFanOut',
+  'largeFanOutThreshold',
+  'crossCuttingRegistries',
+];
+
+export const retirePlanningLimitKnobs = createRetireAgentrcKeyStep({
+  version: '2.57.0',
+  description:
+    'strip the retired planning.* limit knobs from .agentrc.json — ' +
+    'complexityGate, riskHeuristics, the conflict-severity and fan-out ' +
+    'knobs, and memoryPool.{staleAfterDays, growthDelta} (Story #5312)',
+  keys: [
+    ...TOP_LEVEL_KEYS.map((key) => ({
+      path: ['planning', key],
+      pruneDepth: 1,
+    })),
+    { path: ['planning', 'memoryPool', 'staleAfterDays'], pruneDepth: 2 },
+    { path: ['planning', 'memoryPool', 'growthDelta'], pruneDepth: 2 },
+  ],
+});

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mandrel",
-  "version": "2.56.0",
+  "version": "2.57.0",
   "description": "Claude Code-first opinionated workflow framework: instructions, skills, rules, and SDLC workflows that govern AI coding assistants.",
   "files": [
     ".agents/",

package/.agents/scripts/lib/framework-version.js

@@ -1,39 +0,0 @@
-// .agents/scripts/lib/framework-version.js
-/**
- * framework-version.js — the visible authoring-marker surface for the legacy
- * ticket-body provenance stamp.
- *
- * Tickets authored under v1 carry a hybrid stamp: a hidden
- * `<!-- meta: {"mandrel_version":"…","authored_at":"…"} -->` block plus a
- * single visible footer line
- * `> 🏷️ Authored with Mandrel v<version> · <YYYY-MM-DD>`. The producer of
- * new stamps was retired with the Epic write surface (#4574) — nothing stamps
- * a new ticket — but bodies already stamped are live in the backlog, so the
- * Story-body serializer/parser must keep round-tripping them:
- *
- * - {@link AUTHORED_MARKER_LINE_RE} lets the parser skip the marker line during
- *   section parsing so it never pollutes the last structured section.
- * - {@link authoredMarkerLine} lets the serializer re-emit a byte-identical
- *   marker line for a stamp it parsed, preserving provenance verbatim.
- *
- * This module imports nothing so it can be pulled in from the story-body
- * serializer without risking an import cycle.
- */
-
-/**
- * The visible authoring marker line. A blockquote so GitHub renders it as a
- * callout. Used in the Story-body parser to skip the line during section
- * parsing so it never pollutes the last structured section.
- */
-export const AUTHORED_MARKER_LINE_RE = /^\s*>\s*🏷️\s+Authored with Mandrel\b/;
-
-/**
- * Build the visible authoring marker line for a given stamp. The Story-body
- * serializer uses this to re-emit a legacy stamp it parsed, byte-identically.
- *
- * @param {{ version: string, authoredAt: string }} stamp
- * @returns {string}
- */
-export function authoredMarkerLine({ version, authoredAt }) {
-  return `> 🏷️ Authored with Mandrel v${version} · ${authoredAt}`;
-}

package/.agents/scripts/lib/orchestration/consolidation-precondition.js

@@ -1,223 +0,0 @@
-/**
- * consolidation-precondition.js — deterministic dispatch gate for the Phase
- * 8.3 Holistic Consolidation sub-agent (Story #4431, Epic #4429).
- *
- * The Phase 8.3 consolidation critic (`epic-plan-consolidate`) is a genuine
- * fresh-context `Agent` dispatch — every call re-pays the full always-loaded
- * context (`.agents/instructions.md` and its always-on rules, § 4). When the
- * decomposer's draft `tickets.json` already matches the Tech Spec's `##
- * Delivery Slicing` target 1:1 (same shippable-Story count, and the
- * `depends_on` shape already agrees with each slice's declared
- * "Independent?" answer), there is nothing left for the critic to
- * reconcile — dispatching it is pure token spend for a no-op. This module
- * computes that decision **deterministically**, off the same two inputs the
- * critic itself reads (the draft array and the Epic body's Delivery Slicing
- * table), so the planning workflow can skip the sub-agent
- * dispatch when it is provably safe to.
- *
- * **Fail-open by design.** Every ambiguous case — a missing or unparseable
- * Delivery Slicing section, an unparseable "Independent?" cell — resolves to
- * `dispatch: true`. This gate can only ever *save* a dispatch when it is
- * confident the critic has nothing to do; it never disables the critic's
- * ability to catch a real divergence. Since Epic #4474 PR6 this precondition
- * is one input to the risk/size-conditional dispatch layer
- * (`plan-critic-conditions.js`): reachability (8.4) is a deterministic
- * persist-side check (`plan-reachability.js`) and the pre-mortem critic
- * (8.5) is risk/size-gated; the deterministic ticket validator remains
- * unconditional.
- *
- * Pure, synchronous, no I/O — callers own reading `tickets.json` and the
- * Epic body off disk / the GitHub API.
- */
-
-import { DELIVERY_SLICING_RE as DELIVERY_SLICING_HEADING_RE } from '../ticket-body-sections.js';
-
-/** A row is a markdown table line: starts with `|` once trimmed. */
-const TABLE_ROW_RE = /^\|/;
-
-/** A markdown table separator row: `|---|:---:|---:|` (dashes, colons, pipes only). */
-const TABLE_SEPARATOR_RE = /^\|?\s*:?-+:?\s*(\|\s*:?-+:?\s*)*\|?$/;
-
-/** The literal goal-section token that marks the wave-0 BDD scaffold Story. */
-const BDD_SCAFFOLD_GOAL_TOKEN = 'bdd-scaffold';
-
-/**
- * Split one markdown table row into trimmed cell strings.
- *
- * @param {string} line
- * @returns {string[]}
- */
-function splitTableRow(line) {
-  let trimmed = line.trim();
-  if (trimmed.startsWith('|')) trimmed = trimmed.slice(1);
-  if (trimmed.endsWith('|')) trimmed = trimmed.slice(0, -1);
-  return trimmed.split('|').map((cell) => cell.trim());
-}
-
-/**
- * Parse an "Independent?" cell per the pinned rule: match the cell's
- * leading word case-insensitively as `Yes` or `No`. Any other leading word
- * (or an empty cell) is unparseable and returns `null` — the caller must
- * fail open (`dispatch: true`) rather than guess.
- *
- * @param {string} cell
- * @returns {boolean|null} `true` for Yes, `false` for No, `null` when unparseable.
- */
-function parseIndependentCell(cell) {
-  const match = String(cell ?? '')
-    .trim()
-    .match(/^[A-Za-z]+/);
-  if (!match) return null;
-  const word = match[0].toLowerCase();
-  if (word === 'yes') return true;
-  if (word === 'no') return false;
-  return null;
-}
-
-/**
- * Locate and parse the `## Delivery Slicing` markdown table out of the Epic
- * body (which carries the folded Tech Spec sections — Story #4324). Returns
- * `null` when the heading is absent, no table follows it, the table has no
- * "Independent?" column, or any data row's "Independent?" cell is
- * unparseable — every one of those is a fail-open signal for the caller.
- *
- * @param {string} epicBody
- * @returns {{ slice: string, independent: boolean }[] | null}
- */
-export function parseDeliverySlicingTable(epicBody) {
-  if (typeof epicBody !== 'string' || epicBody.length === 0) return null;
-
-  const lines = epicBody.split(/\r?\n/);
-  const headingIdx = lines.findIndex((line) =>
-    DELIVERY_SLICING_HEADING_RE.test(line.trim()),
-  );
-  if (headingIdx === -1) return null;
-
-  let i = headingIdx + 1;
-  while (i < lines.length && lines[i].trim() === '') i++;
-  if (i >= lines.length || !TABLE_ROW_RE.test(lines[i].trim())) return null;
-
-  const headerCells = splitTableRow(lines[i]);
-  i++;
-  if (i >= lines.length || !TABLE_SEPARATOR_RE.test(lines[i].trim())) {
-    return null;
-  }
-  i++;
-
-  const independentIdx = headerCells.findIndex((cell) =>
-    /independent/i.test(cell),
-  );
-  if (independentIdx === -1) return null;
-
-  const rows = [];
-  while (i < lines.length && TABLE_ROW_RE.test(lines[i].trim())) {
-    const cells = splitTableRow(lines[i]);
-    const independent = parseIndependentCell(cells[independentIdx]);
-    if (independent === null) return null; // unparseable cell → fail open
-    rows.push({ slice: (cells[0] ?? '').trim(), independent });
-    i++;
-  }
-
-  return rows.length > 0 ? rows : null;
-}
-
-/**
- * True when `story` is the recognized wave-0 BDD scaffold Story — identified
- * by the literal `bdd-scaffold` goal token the decomposer prompt
- * skill's WAVE-0 BDD SCAFFOLD STORY section requires. Scaffold Stories are
- * not a Delivery Slicing slice, so they are excluded from the count
- * comparison — BDD-adopting consumer repos still benefit from the
- * precondition gate rather than always paying the 8.3 dispatch.
- *
- * @param {{ body?: unknown }} story
- * @returns {boolean}
- */
-function isBddScaffoldStory(story) {
-  const body = story?.body;
-  return (
-    typeof body === 'string' &&
-    body.toLowerCase().includes(BDD_SCAFFOLD_GOAL_TOKEN)
-  );
-}
-
-/**
- * Evaluate whether the Phase 8.3 consolidation sub-agent needs to run.
- *
- * @param {object} input
- * @param {object[]} input.draftStories - The draft `tickets.json` array
- *   (the decomposer's output) — raw Story ticket objects with
- *   top-level `slug` / `depends_on` / `body` (serialized string).
- * @param {string} input.epicBody - The Epic body carrying the folded Tech
- *   Spec sections (`## Delivery Slicing` onward).
- * @returns {{ dispatch: boolean, cause: 'match'|'divergence'|'fail-open', reasons: string[] }}
- *   `dispatch: false` only when the draft matches the Delivery Slicing table
- *   1:1 in count and dependency shape; `dispatch: true` (with `reasons`)
- *   otherwise, including every fail-open case. `cause` distinguishes a
- *   **confirmed** divergence (count or dependency-shape mismatch) from the
- *   fail-open ambiguity (missing/unparseable table) — the #4474 PR6
- *   conditional-dispatch layer treats only the former as a firing condition
- *   on small drafts.
- */
-export function evaluateConsolidationPrecondition({ draftStories, epicBody }) {
-  if (!Array.isArray(draftStories)) {
-    throw new TypeError(
-      'evaluateConsolidationPrecondition: draftStories must be an array',
-    );
-  }
-
-  const slicing = parseDeliverySlicingTable(epicBody);
-  if (!slicing) {
-    return {
-      dispatch: true,
-      cause: 'fail-open',
-      reasons: [
-        'Delivery Slicing section is missing or unparseable — fail-open to the critic.',
-      ],
-    };
-  }
-
-  const slicedStories = draftStories.filter(
-    (story) => !isBddScaffoldStory(story),
-  );
-
-  if (slicedStories.length !== slicing.length) {
-    return {
-      dispatch: true,
-      cause: 'divergence',
-      reasons: [
-        `Story count diverges from Delivery Slicing: ${slicing.length} proposed slice(s) vs ${slicedStories.length} non-scaffold draft Story(ies).`,
-      ],
-    };
-  }
-
-  const reasons = [];
-  for (let idx = 0; idx < slicing.length; idx++) {
-    const slice = slicing[idx];
-    const story = slicedStories[idx];
-    const dependsOn = Array.isArray(story?.depends_on) ? story.depends_on : [];
-    const hasDeps = dependsOn.length > 0;
-    const storyLabel = story?.slug ?? story?.title ?? `<story ${idx + 1}>`;
-
-    if (slice.independent === false && !hasDeps) {
-      reasons.push(
-        `Slice "${slice.slice}" (position ${idx + 1}) is marked Independent: No but draft Story "${storyLabel}" declares no depends_on.`,
-      );
-    } else if (slice.independent === true && hasDeps) {
-      reasons.push(
-        `Slice "${slice.slice}" (position ${idx + 1}) is marked Independent: Yes but draft Story "${storyLabel}" declares depends_on [${dependsOn.join(', ')}].`,
-      );
-    }
-  }
-
-  if (reasons.length > 0) {
-    return { dispatch: true, cause: 'divergence', reasons };
-  }
-
-  return {
-    dispatch: false,
-    cause: 'match',
-    reasons: [
-      `Draft matches Delivery Slicing 1:1 in count and dependency shape (${slicing.length} slice(s)) — skipping the 8.3 consolidation dispatch.`,
-    ],
-  };
-}