mandrel 1.80.0 → 1.82.0

package/.agents/scripts/lib/orchestration/ticket-validator-sizing.js

@@ -86,6 +86,31 @@ function resolveAcceptance(story) {
   return Array.isArray(body?.acceptance) ? body.acceptance : [];
 }
 
+/**
+ * Resolve the `depends_on` edge list for a Story. `depends_on` may live at the
+ * top level (the validator normalizes it there — see `ticket-validator.js`) or
+ * inside the structured body (the shape `story-body.js` parses out). Prefer the
+ * top-level array, falling back to the resolved body's `depends_on`. Returns
+ * only non-empty string slugs — mirroring the `story-body.js` normalization so
+ * a blank / non-string edge never counts as a real dependency.
+ *
+ * The cross-Story validator already rejects any `depends_on` slug that does not
+ * match a sibling in the same decomposition, so every edge that survives to the
+ * sizing layer is guaranteed to point at a sibling. The merge-candidate
+ * heuristic therefore only needs to know whether the list is non-empty.
+ *
+ * @param {object} story
+ * @returns {string[]}
+ */
+function resolveDependsOn(story) {
+  const raw = Array.isArray(story?.depends_on)
+    ? story.depends_on
+    : (resolveStoryBody(story)?.depends_on ?? []);
+  return Array.isArray(raw)
+    ? raw.filter((d) => typeof d === 'string' && d.trim().length > 0)
+    : [];
+}
+
 export const DEFAULT_TASK_SIZING = Object.freeze({
   // Typical-Story warning thresholds (soft — emit advisory findings).
   // Story #4162 raised `softFiles` 8 → 15: a capability-sized Story routinely
@@ -97,6 +122,15 @@ export const DEFAULT_TASK_SIZING = Object.freeze({
   // Hard ceilings (rejection unless lifted).
   hardFiles: 30,
   maxAcceptance: 14,
+  // Under-size (merge-candidate) thresholds (Story #4312). A Story with a
+  // footprint at or below BOTH ceilings that also carries at least one
+  // `depends_on` edge to a sibling looks like a dependent fragment rather than
+  // a capability slice — the machine-checkable form of the single-consumer
+  // merge rule. Emitted as a `soft` advisory only; never a rejection. A tiny
+  // ORPHAN Story (no `depends_on`) stays silent — small orthogonal slices are
+  // legitimate.
+  mergeCandidateMaxFiles: 3,
+  mergeCandidateMaxAcceptance: 4,
 });
 
 /**
@@ -117,6 +151,16 @@ export const DEFAULT_TASK_SIZING = Object.freeze({
  * as a single PR — not a single module or file. Module-level slices fold into
  * the capability they belong to, and a Story whose only consumer is one
  * sibling Story is merged into that sibling (single-consumer merge rule).
+ *
+ * The `envelopeFloor` sentence is the **soft** complement to the sizing
+ * section's envelope framing (Story #4313): the delivery envelope
+ * (`maxTokenBudget`) bounds the TOP of a Story, but under-utilizing it is
+ * itself a merge signal. It is deliberately prose-only — an illustrative
+ * fraction, not a threshold constant, and no validator finding backs it (the
+ * mechanical backstop is the `merge-candidate` finding). It names the
+ * per-Story delivery-session cost (hydration, branch, PR, review, CI) so the
+ * "models one-shot bigger things now" planning assumption is stated, not
+ * implicit.
  */
 export const DELIVERABLE_GRANULARITY_GUIDANCE = Object.freeze({
   // The one-sentence definition of Story granularity.
@@ -125,6 +169,9 @@ export const DELIVERABLE_GRANULARITY_GUIDANCE = Object.freeze({
   // The single-consumer merge rule.
   singleConsumerRule:
     '**Single-consumer merge rule.** A Story whose only consumer is one sibling Story should be **merged into that sibling** rather than emitted separately — a single-consumer downstream slice is not its own unit of work.',
+  // The envelope-floor heuristic (soft; no validator finding — Story #4313).
+  envelopeFloor:
+    '**Envelope floor — under-utilizing the envelope is a merge signal.** A Story that would plausibly consume well under a third of the delivery envelope (`maxTokenBudget`) and is neither parallel-deliverable nor orthogonal to its siblings should be **merged into its consumer**. The fraction is illustrative, not a threshold — the point is that modern frontier models one-shot capability-sized changes, so a chain of small dependent Stories needlessly pays a full delivery session (hydration, branch, PR, review, CI) per link. Merge such links up unless a parallelism or orthogonality reason justifies the separate slice.',
 });
 
 /**
@@ -291,6 +338,66 @@ function computeMissingReasonToExistFinding(story) {
   return hasReason ? [] : [makeMissingReasonToExist(story.slug)];
 }
 
+/**
+ * Soft, advisory `merge-candidate` finding (Story #4312). Surfaces a Story that
+ * looks like a dependent fragment rather than a capability slice — the
+ * machine-checkable form of the single-consumer merge rule
+ * (`DELIVERABLE_GRANULARITY_GUIDANCE.singleConsumerRule`). Never a rejection:
+ * `severity: 'soft'`, so it rides the advisory `findings[]` channel and never
+ * enters the hard `errors[]` array.
+ *
+ * The rendered message names the depended-on sibling slug(s) and recommends
+ * merging into the consumer, mirroring the tone of the `wide-undeclared` nudge.
+ */
+function makeMergeCandidate(slug, fileCount, acceptanceCount, dependsOn) {
+  const siblings = dependsOn.map((d) => `"${d}"`).join(', ');
+  return {
+    kind: 'merge-candidate',
+    severity: 'soft',
+    ticketSlug: slug,
+    fileCount,
+    acceptanceCount,
+    dependsOn,
+    message: `Story "${slug}" is a thin dependent slice (${fileCount} declared file(s), ${acceptanceCount} acceptance item(s)) that depends on sibling(s) ${siblings}. A Story whose only role is to feed one sibling is not its own unit of work — consider merging it into the consumer (single-consumer merge rule) rather than shipping it as a separate slice.`,
+  };
+}
+
+/**
+ * Emit a soft `merge-candidate` finding when a Story meets the under-size
+ * heuristic (Story #4312): footprint ≤ `mergeCandidateMaxFiles` declared
+ * `changes[]` files AND ≤ `mergeCandidateMaxAcceptance` acceptance items AND at
+ * least one `depends_on` edge to a sibling. All three conditions MUST hold — a
+ * tiny ORPHAN Story (no `depends_on`) stays silent because small orthogonal
+ * slices are legitimate. Glob entries mark the footprint as unknown-width, so a
+ * glob-carrying Story is never a merge candidate.
+ *
+ * @param {object} story
+ * @param {{ fileCount: number, hasGlobs: boolean }} changesAnalysis
+ * @param {number} acceptanceCount
+ * @param {object} sizing
+ * @returns {object[]}
+ */
+function computeMergeCandidateFinding(
+  story,
+  changesAnalysis,
+  acceptanceCount,
+  sizing,
+) {
+  if (changesAnalysis.hasGlobs) return [];
+  const dependsOn = resolveDependsOn(story);
+  if (dependsOn.length === 0) return [];
+  if (changesAnalysis.fileCount > sizing.mergeCandidateMaxFiles) return [];
+  if (acceptanceCount > sizing.mergeCandidateMaxAcceptance) return [];
+  return [
+    makeMergeCandidate(
+      story.slug,
+      changesAnalysis.fileCount,
+      acceptanceCount,
+      dependsOn,
+    ),
+  ];
+}
+
 /**
  * Returns true when a `changes[]` entry is a glob pattern. Handles both the
  * canonical PathEntry object form `{ path, assumption }` and legacy strings.
@@ -394,6 +501,7 @@ function computeStorySizingFindings(story, sizing) {
   const acceptance = resolveAcceptance(story);
   const changes = Array.isArray(body?.changes) ? body.changes : [];
   const declaredWide = isDeclaredWide(body?.wide ?? null);
+  const changesAnalysis = analyseChanges(changes);
 
   // Soft, advisory: flag acceptance criteria that reference a configuration
   // constant without inlining a concrete value (Story #3855). Independent of
@@ -407,6 +515,19 @@ function computeStorySizingFindings(story, sizing) {
   // Independent of the numeric sizing layers below.
   out.push(...computeMissingReasonToExistFinding(story));
 
+  // Soft, advisory: flag a thin dependent slice (Story #4312) — the symmetric
+  // under-size backstop to the over-size ceilings. A tiny footprint with a
+  // `depends_on` edge to a sibling is a merge candidate under the
+  // single-consumer merge rule. Independent of the numeric sizing layers below.
+  out.push(
+    ...computeMergeCandidateFinding(
+      story,
+      changesAnalysis,
+      acceptance.length,
+      sizing,
+    ),
+  );
+
   // Acceptance ceiling + soft warn.
   if (acceptance.length > sizing.maxAcceptance) {
     out.push(
@@ -428,7 +549,7 @@ function computeStorySizingFindings(story, sizing) {
     );
   }
 
-  const { fileCount, hasGlobs } = analyseChanges(changes);
+  const { fileCount, hasGlobs } = changesAnalysis;
 
   // Glob entries mark the Story as unknown-width: a glob cannot be bounded by
   // the numeric ceiling, so skip it. A non-wide Story carrying globs gets an

package/.agents/scripts/lib/orchestration/ticketing/bulk.js

@@ -295,15 +295,12 @@ async function processCascadeParentLocked(
     // PR-driven `Closes #N` auto-close, and a recovery transition in
     // `epic-deliver-finalize.js`).
     //
-    // Planning tickets (context::prd, context::tech-spec) DO close via
-    // cascade now (Story #1951). Previously they were excluded under
-    // the assumption that the operator would close them manually
-    // post-merge — but that step never reliably happened and leaving
+    // Legacy planning tickets (pre-#4324 `context::*` artifacts on
+    // historical Epics) DO close via cascade (Story #1951): leaving
     // them open as native sub-issues of the Epic blocks GitHub from
-    // honoring the Epic's `Closes #N` footer. The Epic finalize phase
-    // also closes them explicitly; this cascade branch is the
-    // defense-in-depth path when a Story's tasklist references a
-    // planning ticket directly.
+    // honoring the Epic's `Closes #N` footer. New Epics carry their
+    // planning content on the Epic body itself, so this branch is
+    // purely a legacy-hygiene path.
     //
     // Reuse the parentSnapshot from the idempotency check above — it is
     // a fresh read (cache was invalidated before the getTicket call) and

package/.agents/scripts/lib/orchestration/ticketing/reads.js

@@ -93,7 +93,7 @@ export const STRUCTURED_COMMENT_TYPES = Object.freeze([
   // a sharpened body rewrite, recording the persistence event for audit.
   'clarity-gate-update',
   // Story #2635 — Phase 7 Tech Spec freshness check. `epic-plan-spec.js`
-  // upserts a `spec-freshness` comment on the Tech Spec issue listing any
+  // upserts a `spec-freshness` comment on the Epic listing any
   // path-shaped references that don't exist at the base branch, so the
   // operator can correct drift before Phase 8 decomposes from a stale
   // spec. Advisory: the run continues regardless of the report contents.
@@ -118,7 +118,7 @@ export const STRUCTURED_COMMENT_TYPES = Object.freeze([
   'model-attribution',
   // Story #2894 — `finalize/post-handoff-comment.js` upserts an
   // `epic-handoff` comment on the Epic at the end of the bus-owned
-  // finalize flow (after `open-or-locate-pr` and `close-planning-tickets`
+  // finalize flow (after `open-or-locate-pr`
   // succeed). The marker carries the freshly opened/located PR URL and
   // its number so operators can jump straight to the PR from the Epic
   // ticket. Re-invocations upsert the same marker rather than appending

package/.agents/scripts/lib/plan-phase-cleanup.js

@@ -3,7 +3,7 @@
  *
  * The spec and decompose phases write several Epic-scoped temp files under
  * the per-Epic tree (`temp/epic-<id>/planner-context.json`,
- * `temp/epic-<id>/prd.md`, etc. — see `lib/config/temp-paths.js`). The
+ * `temp/epic-<id>/techspec.md`, etc. — see `lib/config/temp-paths.js`). The
  * workflow .md previously told the operator to `Remove-Item` those files by
  * name at the end of each phase, which rots: adding a new temp file in the
  * script required a synchronized markdown edit, and missed edits left
@@ -37,7 +37,6 @@ import { PROJECT_ROOT, resolveConfig } from './config-resolver.js';
 export const PHASE_TEMP_BASENAMES = Object.freeze({
   spec: Object.freeze([
     'planner-context.json',
-    'prd.md',
     'techspec.md',
     'acceptance-spec.md',
   ]),

package/.agents/scripts/lib/qa/qa-context-hydrator.js

@@ -3,8 +3,9 @@
  *
  * Both QA front-ends (`/qa-explore` and `/qa-run`) need to load the
  * *grounded* surface context for an Epic before they reason about what to test:
- * the Epic body, its linked context tickets (PRD / Tech Spec / Acceptance
- * Spec), the project's `.feature` files, the implementation files the surface
+ * the Epic body (which carries the folded Tech Spec sections and
+ * Acceptance Table — Story #4324 retired the separate context tickets),
+ * the project's `.feature` files, the implementation files the surface
  * map points at, and a slice of recent git history. Today a front-end that
  * trusts in-code comments ("this handler lives at …") can be wrong — the path
  * may have moved, or never existed on the base branch at all. This hydrator
@@ -37,17 +38,6 @@ export const DEFAULT_BASE_REF = 'main';
 /** How many recent commits the hydrator pulls into the context object. */
 export const DEFAULT_LOG_MAX_COUNT = 20;
 
-/**
- * The GitHub label prefixes that mark a linked context ticket. The hydrator
- * resolves `context::prd`, `context::tech-spec`, and `context::acceptance-spec`
- * references off the Epic body and fetches each one through the port.
- */
-export const CONTEXT_TICKET_KINDS = Object.freeze([
-  'prd',
-  'tech-spec',
-  'acceptance-spec',
-]);
-
 /**
  * @typedef {object} GithubPort
  * @property {(issueNumber: number) => Promise<{
@@ -73,57 +63,6 @@ export const CONTEXT_TICKET_KINDS = Object.freeze([
  * @property {string} [note] Free-form provenance note (e.g. a code comment).
  */
 
-/**
- * Parse the linked context-ticket issue numbers out of an Epic body.
- *
- * Two reference shapes are recognized, both written by the planning workflows:
- *   - A labelled line: `context::prd #3800` (or `context::tech-spec: #3801`).
- *   - A "Planning Artifacts" link: `- PRD: #3800`, `- Tech Spec: #3801`,
- *     `- Acceptance Spec: #3802`.
- *
- * Returns a map of kind → issue number for whichever kinds are present. A kind
- * that appears more than once keeps the first occurrence; an unparseable line
- * is skipped rather than thrown.
- *
- * @param {string} epicBody
- * @returns {Record<string, number>}
- */
-export function parseContextTicketRefs(epicBody) {
-  const refs = {};
-  const body = typeof epicBody === 'string' ? epicBody : '';
-
-  // Map the human-readable labels back to canonical kinds.
-  const labelToKind = {
-    prd: 'prd',
-    'tech-spec': 'tech-spec',
-    'tech spec': 'tech-spec',
-    techspec: 'tech-spec',
-    'acceptance-spec': 'acceptance-spec',
-    'acceptance spec': 'acceptance-spec',
-    acceptancespec: 'acceptance-spec',
-  };
-
-  for (const rawLine of body.split('\n')) {
-    const line = rawLine.trim();
-    // Match `context::tech-spec #N`, `Tech Spec: #N`, `- PRD: #N`, etc.
-    const match = line.match(
-      /(?:context::)?([A-Za-z][A-Za-z -]*?)\s*[:#]*\s*#(\d+)/,
-    );
-    if (!match) continue;
-    const key = match[1]
-      .trim()
-      .toLowerCase()
-      .replace(/^context::/, '');
-    const kind = labelToKind[key];
-    if (!kind) continue;
-    const issueNumber = Number(match[2]);
-    if (!Number.isInteger(issueNumber)) continue;
-    if (refs[kind] === undefined) refs[kind] = issueNumber;
-  }
-
-  return refs;
-}
-
 /**
  * Enumerate the `.feature` files under `featureRoot`, returning repo-relative
  * POSIX-style paths sorted for determinism. A missing root yields an empty
@@ -189,9 +128,9 @@ export async function verifySurfaceMap(surfaceMap, gitPort, baseRef) {
  * Hydrate the QA context object for an Epic.
  *
  * Assembles, in one object:
- *   - `epic`            — the Epic's `{ number, body, labels }`.
- *   - `contextTickets`  — the linked PRD / Tech Spec / Acceptance Spec tickets,
- *                         keyed by kind, each `{ number, body, labels }`.
+ *   - `epic`            — the Epic's `{ number, body, labels }`. The body is
+ *                         the single planning document (ideation sections +
+ *                         folded Tech Spec sections + Acceptance Table).
  *   - `featureFiles`    — repo-relative paths of the project's `.feature` files.
  *   - `implementation`  — the verified surface map (each entry carries
  *                         `verified` against the base ref).
@@ -214,9 +153,6 @@ export async function verifySurfaceMap(surfaceMap, gitPort, baseRef) {
  * }} opts
  * @returns {Promise<{
  *   epic: { number: number, body: string, labels: string[] },
- *   contextTickets: Record<string, {
- *     number: number, body: string, labels: string[],
- *   }>,
  *   featureFiles: string[],
  *   implementation: Array<{ path: string, note: string | null, verified: boolean }>,
  *   gitLog: Array<{ sha: string, subject: string }>,
@@ -263,20 +199,6 @@ export async function hydrateQaContext(opts) {
     labels: Array.isArray(epicIssue.labels) ? [...epicIssue.labels] : [],
   };
 
-  // Resolve and fetch the linked context tickets off the Epic body.
-  const refs = parseContextTicketRefs(epic.body);
-  const contextTickets = {};
-  for (const kind of CONTEXT_TICKET_KINDS) {
-    const number = refs[kind];
-    if (number === undefined) continue;
-    const issue = await githubPort.fetchIssue(number);
-    contextTickets[kind] = {
-      number: issue.number,
-      body: issue.body ?? '',
-      labels: Array.isArray(issue.labels) ? [...issue.labels] : [],
-    };
-  }
-
   const featureFiles = collectFeatureFiles(featureRoot, { fsImpl });
   const implementation = await verifySurfaceMap(surfaceMap, gitPort, baseRef);
   const gitLog = await gitPort.recentLog({ maxCount: logMaxCount });
@@ -286,7 +208,6 @@ export async function hydrateQaContext(opts) {
 
   return {
     epic,
-    contextTickets,
     featureFiles,
     implementation,
     gitLog: Array.isArray(gitLog) ? gitLog : [],

package/.agents/scripts/lib/templates/decomposer-prompts.js

@@ -52,12 +52,16 @@ function render2TierPrompt({ maxTickets, maxTokenBudget, epicId = null }) {
   // (ticket-validator-sizing.js) so the prompt and the validator cannot drift.
   const { softFiles, hardFiles, maxAcceptance, softAcceptanceCount } =
     DEFAULT_TASK_SIZING;
-  // Deliverable-granularity definition + single-consumer merge rule are
-  // sourced from the single DELIVERABLE_GRANULARITY_GUIDANCE constant
-  // (ticket-validator-sizing.js) so the prompt and the authoring SKILL
-  // cannot drift (Story #3777).
-  const { definition: granularityDefinition, singleConsumerRule } =
-    DELIVERABLE_GRANULARITY_GUIDANCE;
+  // Deliverable-granularity definition + single-consumer merge rule + the
+  // soft envelope-floor heuristic are sourced from the single
+  // DELIVERABLE_GRANULARITY_GUIDANCE constant (ticket-validator-sizing.js) so
+  // the prompt and the authoring SKILL cannot drift (Story #3777; the
+  // envelope-floor sentence added by Story #4313).
+  const {
+    definition: granularityDefinition,
+    singleConsumerRule,
+    envelopeFloor,
+  } = DELIVERABLE_GRANULARITY_GUIDANCE;
   // The binding-vs-advisory authoring altitude + the New-File Contract are
   // sourced from the single AUTHORING_ALTITUDE_GUIDANCE constant
   // (ticket-validator-sizing.js) so the prompt and the authoring SKILL cannot
@@ -75,7 +79,7 @@ function render2TierPrompt({ maxTickets, maxTokenBudget, epicId = null }) {
     ? `@epic-${epicId}-ac-1`
     : '@epic-<id>-ac-N';
   return `You are an expert Senior Project Manager and Orchestrator.
-Your job is to take a Product Requirements Document (PRD) and a Technical Specification and decompose them into a flat list of Story tickets for an AI Agent to execute.
+Your job is to take an Epic (including its inline User Stories) and a Technical Specification and decompose them into a flat list of Story tickets for an AI Agent to execute.
 
 ### HIERARCHY RULES:
 1. **Stories**: Specific user-facing or architectural user stories (e.g., "Implement JWT Token Exchange").
@@ -162,6 +166,8 @@ The primary question is **cohesion, not count**: *is this one coherent change wi
 
 **Size against the real one-pass delivery envelope.** Each Story is delivered and self-verified by a single agent in one pass, whose context is capped by the delivery token budget \`maxTokenBudget = ${maxTokenBudget}\` tokens (the task-prompt hydration cap). Use that envelope — not the file count alone — as the leading sizing input: a Story is correctly sized when one agent can hold its full change, acceptance, and verification in a single pass within \`maxTokenBudget\`. The numeric file thresholds below are a coarse backstop on top of this envelope, not the primary signal.
 
+${envelopeFloor}
+
 - **One Story = one coherent change with one reason to exist.** If you cannot state that reason in a sentence, the Story is probably two Stories — or two Stories that should be one. State that sentence explicitly in the Story's \`reason_to_exist\` meta field (see STORY BODY RULES) so the consolidate critic can check it.
 - ${singleConsumerRule}
 - **Split independent, parallelizable work** into sibling Stories — but only when the pieces genuinely have separate reasons to exist.
@@ -194,7 +200,7 @@ Declaring \`wide\` with a non-empty reason **lifts the \`hardFiles\` rejection**
 
 #### BRAND / COPY / STYLE WORK:
 
-- Stories that touch user-visible copy, brand assets, or visual style MUST cite the relevant section of \`docs/style-guide.md\` in \`acceptance\` (e.g. \`"acceptance": ["Hero copy matches docs/style-guide.md §3 (voice & tone)"]\`). If \`docs/style-guide.md\` does not exist or has no relevant section, state that explicitly: \`"acceptance": ["docs/style-guide.md absent — copy reviewed against the inline brand brief in PRD §2"]\`. Silence on style sourcing is a smell.
+- Stories that touch user-visible copy, brand assets, or visual style MUST cite the relevant section of \`docs/style-guide.md\` in \`acceptance\` (e.g. \`"acceptance": ["Hero copy matches docs/style-guide.md §3 (voice & tone)"]\`). If \`docs/style-guide.md\` does not exist or has no relevant section, state that explicitly: \`"acceptance": ["docs/style-guide.md absent — copy reviewed against the inline brand brief in the Epic body"]\`. Silence on style sourcing is a smell.
 
 ### WAVE-0 BDD SCAFFOLD STORY (features-first; emit when the Acceptance Spec has \`new\`-disposition rows):
 The Acceptance Spec's AC table (columns \`AC ID | Outcome | Feature File | Scenario | Disposition\`) tags each row's \`Disposition\` with one of \`new | updated | unchanged\`. A \`new\` row names a \`.feature\` file + scenario that does NOT yet exist on \`main\`. The framework is features-first: implementing Stories reference those \`.feature\` paths in their \`verify[]\` lines, so the files MUST already exist when those Stories run — otherwise verification fails mid-delivery on a missing file. (These Gherkin \`.feature\` files are BDD artifacts, unrelated to any ticket tier.)

package/.agents/scripts/lifecycle-emit.js

@@ -412,7 +412,7 @@ export async function runLifecycleEmit({
   // record `failed` classifications instead of throwing (the bus's
   // `onFailed` boundary already persists the originating event), so the
   // bus emit resolves cleanly even when a downstream side effect — e.g.
-  // an acceptance-reconcile gap or a `closePlanningTickets` throw inside
+  // an acceptance-reconcile gap or an `openOrLocatePr` throw inside
   // the Finalizer — failed. Without surfacing these, the CLI would exit 0
   // with success-shaped JSON despite a partial finalize (Story #3904).
   const outcomes = collectOutcomes(chain);

package/.agents/scripts/lint-label-vocabulary.js

@@ -66,7 +66,6 @@ export const KNOWN_AXES = Object.freeze([
   'type',
   'status',
   'persona',
-  'context',
   'acceptance',
   'meta',
   'planning',
@@ -156,11 +155,11 @@ export function findVocabularyViolations(src, axes = KNOWN_AXES) {
   //   - value is a lowercase identifier (`[a-z][a-z0-9-]*`)
   //
   // The whole-body anchor avoids false positives on:
-  //   - `context::acceptance-spec` — canonical label, value contains a
+  //   - `acceptance::n-a` — canonical label, value contains a
   //     hyphen but is not at axis position.
   //   - `<type>/<slug>` — template placeholders.
   //   - `delivery.maxTokenBudget` — config-key paths with `.`.
-  //   - `context.{prdId,techSpecId}` — JS destructure shapes.
+  //   - `delivery.{maxTickets}` — JS destructure shapes.
   //   - `agent-protocol.md` — filenames / concept slugs.
   //
   // Real drift like `type/epic` matches because the WHOLE body is

package/.agents/scripts/providers/github/mappers.js

@@ -11,8 +11,6 @@
  * Extracted from `../github.js` in Story #1846 / Task #1859.
  */
 
-import { parseLinkedIssues } from '../../lib/issue-link-parser.js';
-
 function normalizeLabels(issue) {
   const raw = issue?.labels;
   if (!raw) return [];
@@ -50,7 +48,6 @@ export function issueToEpic(issue) {
     body: issue.body ?? '',
     labels,
     labelSet: new Set(labels),
-    linkedIssues: parseLinkedIssues(issue.body),
   };
 }
 

package/.agents/scripts/providers/github/tickets.js

@@ -112,7 +112,7 @@ export class TicketGateway {
     this._cache = cache ?? createInlineTicketCache();
     /**
      * Per-instance memo of `getTickets(epicId, filters)` results (Story
-     * #3988). The planning-state-manager fetches the same child list twice
+     * #3988). Planning-era healing fetched the same child list twice
      * per planning pass; without this memo each fetch re-pays the full
      * search/list round-trip. Invalidated on every write surface.
      * @type {Map<string, object[]>}
@@ -322,24 +322,13 @@ export class TicketGateway {
     // Mirror the Epic create path (issues.js:160 → `labels: TYPE_LABELS.EPIC`):
     // inject TYPE_LABELS.STORY so a spec that omits the labels array cannot
     // produce an unlabeled, undispatchable Story. Dedupe to avoid duplicates
-    // when the caller already carries the label.
-    //
-    // Context spec tickets (PRD / Tech Spec / Acceptance Spec) are a distinct
-    // ticket class created through this same factory carrying a `context::*`
-    // label (and no `type::story`). They MUST NOT be stamped `type::story`:
-    // doing so makes every `type::story`-counting consumer — the decompose
-    // open-children guard (`assertNoOpenPlanChildren`), the delivery wave
-    // builder (`discoverOpenStories`), the plan healthcheck — mis-classify
-    // them as deliverable Stories. Skip the injection when the caller's labels
-    // already classify the ticket as context.
+    // when the caller already carries the label. (Story #4324 retired the
+    // `context::*` ticket classes, so every ticket created through this
+    // factory is a Story.)
     const callerLabels = ticketData.labels ?? [];
-    const isContextTicket = callerLabels.some(
-      (l) => typeof l === 'string' && l.startsWith('context::'),
-    );
-    const labels =
-      isContextTicket || callerLabels.includes(TYPE_LABELS.STORY)
-        ? callerLabels
-        : [TYPE_LABELS.STORY, ...callerLabels];
+    const labels = callerLabels.includes(TYPE_LABELS.STORY)
+      ? callerLabels
+      : [TYPE_LABELS.STORY, ...callerLabels];
     const result = await this._gh.api({
       method: 'POST',
       endpoint: `/repos/${this.owner}/${this.repo}/issues`,

package/.agents/scripts/single-story-init.js

@@ -23,7 +23,6 @@
  *   6. Flip the Story to `agent::executing`.
  *
  * What this script does NOT do (and why):
- *   - Skips `traceHierarchy` — no Epic → no PRD/Tech-Spec.
  *   - Skips `validateBlockers` against the body's `Blocked by:` markers —
  *     pre-flight is still the operator's responsibility, but the Epic-scope
  *     blocker chain doesn't fit.

package/.agents/scripts/story-init.js

@@ -9,7 +9,6 @@
  * pipeline stages from `lib/story-init/`:
  *
  *   1. context-resolver     — fetch the Story + optionally mark as recut.
- *   2. hierarchy-tracer     — resolve Feature/Epic → PRD / Tech Spec.
  *   3. blocker-validator    — refuse to proceed while dependencies are open.
  *   4. task-graph-builder   — fetch + topologically sort child Tasks.
  *   5. branch-initializer   — materialise the story branch (single-tree
@@ -53,7 +52,6 @@ import {
   planStoryBranchSeed,
 } from './lib/story-init/branch-initializer.js';
 import { resolveContext } from './lib/story-init/context-resolver.js';
-import { traceHierarchy } from './lib/story-init/hierarchy-tracer.js';
 import { transitionStoryToExecuting } from './lib/story-init/state-transitioner.js';
 import { buildTaskGraph } from './lib/story-init/task-graph-builder.js';
 import { createPhaseTimer } from './lib/util/phase-timer.js';
@@ -88,8 +86,6 @@ export async function runStoryInit({
   dryRun: dryRunParam,
   cwd: cwdParam,
   recutOf: recutOfParam,
-  prdId: prdIdParam,
-  techSpecId: techSpecIdParam,
   injectedProvider,
   injectedConfig,
 } = {}) {
@@ -100,17 +96,10 @@ export async function runStoryInit({
           dryRun: !!dryRunParam,
           cwd: cwdParam ?? null,
           recutOf: recutOfParam ?? null,
-          prdId: prdIdParam ?? null,
-          techSpecId: techSpecIdParam ?? null,
         }
       : parseSprintArgs();
   const { storyId, dryRun } = parsed;
   const recutOf = recutOfParam ?? parsed.recutOf ?? null;
-  // Story #4253: pre-resolved Epic linkages (from the /deliver fan-out's
-  // one-time Epic resolution). When both are present, hierarchy-tracer skips
-  // the per-Story getEpic round-trip; when absent it resolves them itself.
-  const threadedPrdId = prdIdParam ?? parsed.prdId ?? null;
-  const threadedTechSpecId = techSpecIdParam ?? parsed.techSpecId ?? null;
   // Worktree-aware cwd resolution: explicit param > --cwd flag > env > PROJECT_ROOT.
   const cwd = path.resolve(cwdParam ?? parsed.cwd ?? PROJECT_ROOT);
 
@@ -155,20 +144,7 @@ export async function runStoryInit({
     input: { storyId, recutOf, dryRun },
   });
 
-  // Stage 2 — hierarchy. When the /deliver fan-out threaded --prd/--tech-spec
-  // (both resolved once by the parent), this short-circuits the per-Story
-  // getEpic. Absent flags fall back to the legacy getEpic resolution.
-  const { prdId, techSpecId } = await traceHierarchy({
-    provider,
-    logger: stageLogger,
-    input: { epicId, prdId: threadedPrdId, techSpecId: threadedTechSpecId },
-  });
-
   progress('CONTEXT', `Epic: #${epicId}, Parent: #${parentId ?? 'none'}`);
-  progress(
-    'CONTEXT',
-    `PRD: #${prdId ?? 'none'}, Tech Spec: #${techSpecId ?? 'none'}`,
-  );
 
   // Stage 3 — blockers.
   const { openBlockers } = await validateBlockers({
@@ -335,8 +311,6 @@ export async function runStoryInit({
     installStatus,
     sortedTasks,
     parentId,
-    prdId,
-    techSpecId,
     dryRun,
     recutOf,
     hierarchy: hierarchyMode,
@@ -473,8 +447,6 @@ function buildStoryInitResult({
   installStatus,
   sortedTasks,
   parentId,
-  prdId,
-  techSpecId,
   dryRun,
   recutOf,
   hierarchy,
@@ -506,7 +478,7 @@ function buildStoryInitResult({
       labels: t.labels,
       dependencies: t.dependsOn ?? parseBlockedBy(t.body ?? ''),
     })),
-    context: { parentId, prdId, techSpecId },
+    context: { parentId },
     dryRun,
   };
 }