create-pmos 0.78.0 → 0.79.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "create-pmos",
-  "version": "0.78.0",
+  "version": "0.79.0",
   "description": "Scaffold the PMOS portable kit — run the PMOS product-management workflow (initiative → lane-fit spec → contract → build → three gates, with work-type lanes for features, research, maintenance, incidents, and content) entirely inside your own repo, with all state as git-versioned files. Includes `workflow-sweep`, which audits the state every board derives from and separates real defects from work you simply have not done. `create-pmos workspace` also installs + launches the local Terminal Workspace GUI. No database, no network, no telemetry.",
   "license": "MIT",
   "bin": {

package/payload/KIT-VERSION

@@ -1,3 +1,3 @@
-kernel-source: pmos@1dbce98
+kernel-source: pmos@4b6af6a
 assembled: 2026-10-07
 mode: vendor

package/payload/planning/evals/eval-calibration.md

@@ -2,7 +2,7 @@
 type: Reference
 title: "Eval calibration log"
 description: "Append-only record of every PM override of an evaluator verdict."
-generated: {by: process:create-pmos, at: "2026-10-07T10:39:13Z"}
+generated: {by: process:create-pmos, at: "2026-10-07T12:01:33Z"}
 status: stable
 ---
 

package/payload/planning/okrs/health-budgets.md

@@ -2,7 +2,7 @@
 type: Reference
 title: "Health budgets"
 description: "Closed registry of health-budget anchors (D59), edited only by the PM."
-generated: {by: process:create-pmos, at: "2026-10-07T10:39:13Z"}
+generated: {by: process:create-pmos, at: "2026-10-07T12:01:33Z"}
 status: stable
 ---
 

package/payload/scripts/workflow-state.js

@@ -720,6 +720,8 @@ function computeWorkflowDerived(id, opts) {
     return Number.isFinite(t) ? t : null;
   })();
   const steps = [];
+  // Steps the PM cut or made optional on 2026-10-07 (planning/initiatives/sdlc-step-cuts.md).
+  const CUT_NOTE = 'cut by the PM 2026-10-07 (sdlc-step-cuts)';
 
   // ── route ─ the lane is chosen and the two intake questions are answered ─────────────────────
   const sRoute = specimenRoute(fm.specimen_test);
@@ -731,6 +733,9 @@ function computeWorkflowDerived(id, opts) {
     if (!missing.length) {
       steps.push(step('route', 'route', 'done',
         `${lane} lane · specimen test ${sRoute} · surface ${fRoute}`));
+    } else if (effLane !== 'feature') {
+      // The routing questions are asked of feature work only.
+      steps.push(step('route', 'route', 'done', `${lane} lane — routing questions are for features only (${CUT_NOTE})`));
     } else if (legacy) {
       steps.push(step('route', 'route', 'unknown',
         `${lane} lane · ${missing.join(' + ')} not declared (record predates ${ENFORCED_FROM})`));
@@ -875,41 +880,21 @@ function computeWorkflowDerived(id, opts) {
     }
   }
 
-  // ── understand ─ O4: an existing surface is explained BEFORE it is specified ─────────────────
+  // ── understand ─ O4, made optional (sdlc-step-cuts) ───────────────────────────────────────
+  // An explain-surface pass that exists is shown; a missing one never blocks.
   {
     const expl = String(fm.explainer_path || '').trim();
-    if (fRoute !== 'existing') {
-      steps.push(step('understand', 'understand', 'na',
-        fRoute === 'greenfield' ? 'greenfield — nothing existing to explain first' : 'surface check not declared'));
-    } else if (!expl) {
-      steps.push(step('understand', 'understand', legacy ? 'unknown' : 'blocked',
-        'surface_check says existing but no explainer_path is set',
-        'run build-skills/explain-surface.skill in a FRESH context (not the agent that will write the spec), then set explainer_path:',
-        { blocking: !legacy, rule: 'O4' }));
-    } else if (!fs.existsSync(path.join(root, expl))) {
-      steps.push(step('understand', 'understand', legacy ? 'unknown' : 'blocked',
-        `explainer_path '${expl}' does not exist in the repo`,
-        'set explainer_path to a real explain-surface output',
-        { blocking: !legacy, rule: 'O4' }));
+    if (expl && fs.existsSync(path.join(root, expl))) {
+      steps.push(step('understand', 'understand', 'done', `${expl} — optional surface explainer`));
     } else {
-      const eIdx = firstAppearance(root, range, expl);
-      const ok = ordered(eIdx, prdIdx);
-      if (ok === null) {
-        steps.push(step('understand', 'understand', 'unknown',
-          `${expl} present; order undecidable (${range.ok ? 'no spec artifact to order against' : range.reason})`));
-      } else if (ok) {
-        steps.push(step('understand', 'understand', 'done', `${expl} — first-appeared before ${prd}`));
-      } else {
-        steps.push(step('understand', 'understand', legacy ? 'unknown' : 'blocked',
-          `ORDER VIOLATION — the spec (${prd}) first-appeared BEFORE the explainer (${expl})`,
-          'without the pass first, the AGENT acquires the understanding and the PM approves a spec built on facts they never saw',
-          { blocking: !legacy, rule: 'O4' }));
-      }
+      steps.push(step('understand', 'understand', 'na', `the surface explainer is optional — ${CUT_NOTE}`));
     }
   }
 
   // ── spec ─ the lane's artifacts, and O1/O2: they precede the build ───────────────────────────
-  const needsRubric = effLane === 'feature';
+  // A feature PRD's done-criteria section can serve as its rubric (sdlc-step-cuts), so a separate
+  // rubric file is optional in every lane. One that is declared must still exist.
+  const needsRubric = false;
   {
     const wants = {
       feature: 'a PRD', discovery: 'a spike brief', incident: 'an incident record',
@@ -925,7 +910,7 @@ function computeWorkflowDerived(id, opts) {
     } else if (!prd) {
       steps.push(step('spec', 'spec', 'blocked', `no prd_path — the ${effLane} lane needs ${wants}`,
         effLane === 'feature'
-          ? 'author the PRD (skills/core/prd.skill) and the rubric alongside it (D13), then set prd_path + rubric_path'
+          ? 'author the PRD (skills/core/prd.skill) with its done-criteria (D13), then set prd_path'
           : `author ${wants} and set prd_path in ${recRel}`,
         { blocking: true, rule: 'R2' }));
     } else if (prdIdx === null) {
@@ -970,88 +955,24 @@ function computeWorkflowDerived(id, opts) {
   const runs = cachedJsonl(path.join(L.runsDir, L.runsFile)).filter((r) => r && r.initiative_id === id);
   const runsRel = path.relative(root, path.join(L.runsDir, L.runsFile));
 
-  // ── priors ─ O7: the episodic read is CITED, and the citation is checked ─────────────────────
-  // AGENTS.md step 3 mandates reading this initiative's earlier runs BEFORE acting — the one kernel
-  // invariant that was still asserted only in prose. Nothing can derive that a file was READ, and
-  // this rule does not pretend to: what it derives is the three things that ARE decidable from the
-  // record, and together they are what separates it from a checkbox.
-  //   1. whether earlier runs existed at all      — the rule is vacuous on a first run, and says so
-  //   2. whether the run declared what it consulted
-  //   3. whether the ids it named are REAL and actually EARLIER
-  // (3) is the part that matters. O5's `PM_approved: true` is a bare boolean, satisfiable by typing
-  // it; a fabricated `priors` list fails against the corpus it claims to have read. The rule is
-  // NON-BLOCKING for the same reason O5/O6 are: it cannot be true before a run exists, and a rule
-  // enforced at a moment it cannot be satisfied only teaches people to reach for the escape hatch.
-  {
-    const latest = runs.length ? runs[runs.length - 1] : null;
-    const earlier = runs.slice(0, -1);
-    const cite = priorsCitation(latest);
-    const how = `consult this initiative's earlier runs BEFORE acting (AGENTS.md step 3) and cite them as \`priors\` on the run in ${runsRel}`;
-    if (!latest) {
-      steps.push(step('priors', 'priors', 'pending', 'no run logged for this initiative yet', how, { rule: 'O7' }));
-    } else if (!earlier.length) {
-      // Not a pass — there was nothing to pass. A tick here would be the watermelon this engine exists to prevent.
-      steps.push(step('priors', 'priors', 'na',
-        `${latest.run_id} is the first run on this initiative — no priors to consult`));
-    } else if (cite.malformed) {
-      steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-        `${latest.run_id} has a malformed priors citation — ${cite.malformed}`,
-        `use \`priors: ["<run_id>", …]\` or \`priors: {consulted: [...], note: "…"}\` in ${runsRel}`,
-        { rule: 'O7' }));
-    } else if (!cite.declared) {
-      steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-        `${latest.run_id} follows ${earlier.length} earlier run(s) on this initiative and cites no priors`,
-        how, { rule: 'O7' }));
-    } else {
-      const byId = new Map(runs.map((r) => [String(r.run_id), r]));
-      const earlierIds = new Set(earlier.map((r) => String(r.run_id)));
-      const missing = cite.consulted.filter((x) => !byId.has(x));
-      const notEarlier = cite.consulted.filter((x) => byId.has(x) && !earlierIds.has(x));
-      if (missing.length) {
-        // The check a declared field would otherwise never get. A citation naming a run that does not
-        // exist reports a read that cannot have happened — worse than an absent citation, because it
-        // reads as compliance on the board.
-        steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-          `${latest.run_id} cites prior run(s) that do not exist on this initiative: ${missing.join(', ')}`,
-          `cite run_ids that are actually in ${runsRel} — a citation naming nothing reports a read that cannot have happened`,
-          { rule: 'O7' }));
-      } else if (notEarlier.length) {
-        steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-          `${latest.run_id} cites ${notEarlier.join(', ')}, which ${notEarlier.length > 1 ? 'are' : 'is'} not EARLIER than it`,
-          'a run cannot consult itself or a run that came after it — cite only runs logged before this one',
-          { rule: 'O7' }));
-      } else if (!cite.consulted.length && !cite.note) {
-        steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-          `${latest.run_id} declares an empty priors list with no note, over ${earlier.length} earlier run(s)`,
-          'an empty list with no reason is indistinguishable from not looking — say which priors were read, or say in `note` why none applied',
-          { rule: 'O7' }));
-      } else if (!cite.consulted.length) {
-        // Allowed, and deliberately not silent: the machine cannot judge whether the priors were
-        // genuinely irrelevant, so it surfaces the claim verbatim for the PM instead of ticking.
-        steps.push(step('priors', 'priors', 'done',
-          `${latest.run_id} consulted none of ${earlier.length} earlier run(s) — declared irrelevant: ${cite.note}`));
-      } else {
-        steps.push(step('priors', 'priors', 'done',
-          `${latest.run_id} cites ${cite.consulted.length} prior run(s): ${cite.consulted.join(', ')}${cite.note ? ` — ${cite.note}` : ''}`));
-      }
-    }
-  }
+  // ── priors ─ O7, cut (sdlc-step-cuts) ───────────────────────────────────────────────────────
+  // Reading earlier runs is still good practice; citing them is no longer checked.
+  steps.push(step('priors', 'priors', 'na', `prior-run citation is not required — ${CUT_NOTE}`));
 
-  // ── contract ─ O5: PM-approved done-criteria before the build. Reported, never CI-blocking ────
+  // ── contract ─ O5: PM-approved done-criteria. Reported, never blocking ─────────────────────
+  // Since sdlc-step-cuts only the feature lane asks for it, and a missing sign-off is pending, not
+  // blocked: the PM's approval of the PRD and the acceptance at the end are the sign-offs that count.
   const approved = runs.find((r) => r.pre_run_contract && r.pre_run_contract.PM_approved === true);
   {
     if (approved) {
       steps.push(step('contract', 'contract', 'done',
         `PM-approved pre_run_contract on run ${approved.run_id}${approved.pre_run_contract.approved_by ? ` (by ${approved.pre_run_contract.approved_by})` : ''}`));
-    } else if (runs.length) {
-      steps.push(step('contract', 'contract', legacy ? 'pending' : 'blocked',
-        `${runs.length} run(s) logged, none with pre_run_contract.PM_approved: true`,
-        `agree numbered done_criteria + test_method with the PM and record the approval in ${runsRel}`,
-        { rule: 'O5' }));
+    } else if (effLane !== 'feature') {
+      steps.push(step('contract', 'contract', 'na', `${effLane} lane — the PM's acceptance stands in for a contract sign-off (${CUT_NOTE})`));
     } else {
       steps.push(step('contract', 'contract', 'pending',
-        'no run logged for this initiative yet',
-        `agree done_criteria + test_method BEFORE building and append the run to ${runsRel} with PM_approved: true`,
+        runs.length ? `${runs.length} run(s) logged, none with pre_run_contract.PM_approved: true` : 'no run logged for this initiative yet',
+        `agree done_criteria + test_method with the PM (approving the PRD counts) and record it in ${runsRel}`,
         { rule: 'O5' }));
     }
   }
@@ -1143,41 +1064,32 @@ function computeWorkflowDerived(id, opts) {
        * fix that made a chore's gate visible ADDED one. The module's own rule, repeated in three
        * files — "a skip is an honest RECORD that verification did not happen; it is not
        * verification" — now applies to the surface that reports it. */
+      // The Review Gate runs on request since sdlc-step-cuts: a skip is n/a, and a FAIL is reported
+      // for the PM to weigh at acceptance rather than holding the gate shut.
       if (verdict === 'SKIPPED') {
-        steps.push(step('review', 'review', 'pending', label,
-          'the gate was skipped, so it is still owed — commission an independent evaluator and record a real verdict',
-          { rule: 'R3' }));
+        steps.push(step('review', 'review', 'na', `${label} — skipped; the Review Gate runs on request (${CUT_NOTE})`));
       } else if (verdict === 'FAIL') {
-        steps.push(step('review', 'review', 'blocked', label,
-          'the Review Gate FAILED — remediate the findings and record a new verdict', { rule: 'R3' }));
+        steps.push(step('review', 'review', 'pending', label,
+          'the Review Gate FAILED — weigh the findings at acceptance, or remediate and record a new verdict', { rule: 'R3' }));
       } else {
         steps.push(step('review', 'review', 'done', label));
       }
     } else {
-      steps.push(step('review', 'review', 'pending', 'no Review-Gate verdict recorded',
-        'score the build against the rubric (build-skills/evaluator.skill), then append to evals.jsonl',
-        { rule: 'R3' }));
+      steps.push(step('review', 'review', 'na', `no Review-Gate verdict — it runs on request (${CUT_NOTE})`));
     }
   }
 
-  // ── explain ─ O6: the run is explained AFTER the build and BEFORE acceptance ─────────────────
+  // ── explain ─ O6, made optional (sdlc-step-cuts) ────────────────────────────────────────────
+  // A run explainer that exists is shown; a missing one never blocks acceptance.
   {
     let found = null;
     for (const r of runs) {
       const p = path.join(L.explainers, `${r.run_id}-explainer.md`);
       if (fs.existsSync(p)) { found = path.relative(root, p); break; }
     }
-    if (found) {
-      steps.push(step('explain', 'explain', 'done', found));
-    } else if (!runs.length) {
-      steps.push(step('explain', 'explain', 'pending', 'no run to explain yet',
-        'log the run first, then run build-skills/explain-run.skill against its diff', { rule: 'O6' }));
-    } else {
-      steps.push(step('explain', 'explain', legacy ? 'pending' : 'blocked',
-        `no explainer at ${path.relative(root, L.explainers)}/${runs[runs.length - 1].run_id}-explainer.md`,
-        'run build-skills/explain-run.skill — raw-transcript vigilance degrades within ~30 min, so the explainer is what the PM actually reviews',
-        { rule: 'O6' }));
-    }
+    steps.push(found
+      ? step('explain', 'explain', 'done', found)
+      : step('explain', 'explain', 'na', `the run explainer is optional — ${CUT_NOTE}`));
   }
 
   // ── accept ─ the human PM, final and non-delegable (D12) ─────────────────────────────────────

package/payload/state/discoveries.md

@@ -2,7 +2,7 @@
 type: Reference
 title: "Discoveries and friction log"
 description: "Append-only record of what runs discovered and where the process rubbed."
-generated: {by: process:create-pmos, at: "2026-10-07T10:39:13Z"}
+generated: {by: process:create-pmos, at: "2026-10-07T12:01:33Z"}
 status: stable
 ---
 

package/workspace/ci-checks.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+# The fast workspace checks, run first in the Workspace CI test job (hosted or self-hosted).
+# Usage: bash ci-checks.sh [BASE_REF]   — run from workspace/. With BASE_REF (a pull request's base
+# branch) the guide-sync and conformance checks also read the PR's diff.
+set -euo pipefail
+base="${1:-}"
+cd "$(dirname "$0")"
+
+if [ -n "$base" ]; then
+  git fetch --no-tags -q origin "$base"
+  # guide-sync: a pull request that changes the workspace front end also changes the guide.
+  changed=$(git -C .. diff --name-only "origin/$base"...HEAD)
+  if echo "$changed" | grep -q '^workspace/public/pm/' && ! echo "$changed" | grep -q '^workspace/guide/'; then
+    echo "❌ workspace/public/pm/** changed but workspace/guide/** did not. Update the guide page that describes this behaviour (AGENTS.md, 'The user guide')."; exit 1
+  fi
+  echo "✅ guide-sync holds."
+fi
+
+# token hygiene: no raw hex outside the token-definition blocks.
+shopt -s nullglob
+files=(public/pm/*.css)
+[ ${#files[@]} -gt 0 ] || { echo "❌ no stylesheet found under public/pm/ — this check would silently pass"; exit 1; }
+hits=$(awk '/^:root\{/,/^\}/{next} /^\[data-pm-theme="light"\]\{/,/^\}/{next} 1' "${files[@]}" | grep -nE '#[0-9a-fA-F]{3,6}' || true)
+[ -z "$hits" ] || { echo "❌ raw hex in a component rule (use a semantic token):"; echo "$hits"; exit 1; }
+echo "✅ token hygiene holds across ${#files[@]} stylesheet(s)."
+
+# conformance gate (R28): every line of new files, the added lines of changed ones.
+node conformance-gate.js --self-test
+[ -z "$base" ] || node conformance-gate.js --base "origin/$base"
+
+# isolation: workspace/ never imports from web/.
+hits=$(grep -rnE "(from|import|require).*['\"][^'\"]*/?web/" server.js public test.js 2>/dev/null || true)
+[ -z "$hits" ] || { echo "❌ workspace/ reaches into web/ (behavior-preservation floor):"; echo "$hits"; exit 1; }
+echo "✅ isolation holds."
+
+# test lists: every suite is listed once and the shards cover them all.
+node run-tests.js --check

package/workspace/fixtures/initiatives-api/roadmap.json

@@ -175,8 +175,15 @@
         "line2": ""
       },
       "health": {
-        "level": "on",
-        "reasons": []
+        "level": "at-risk",
+        "reasons": [
+          {
+            "level": "at-risk",
+            "code": "pm-step-owed",
+            "text": "Your acceptance is owed. The last evidence is 128 d old",
+            "source": "planning/initiatives/fx-old.md"
+          }
+        ]
       },
       "progress": {
         "done": 0,

package/workspace/lib/engine/workflow-state.js

@@ -720,6 +720,8 @@ function computeWorkflowDerived(id, opts) {
     return Number.isFinite(t) ? t : null;
   })();
   const steps = [];
+  // Steps the PM cut or made optional on 2026-10-07 (planning/initiatives/sdlc-step-cuts.md).
+  const CUT_NOTE = 'cut by the PM 2026-10-07 (sdlc-step-cuts)';
 
   // ── route ─ the lane is chosen and the two intake questions are answered ─────────────────────
   const sRoute = specimenRoute(fm.specimen_test);
@@ -731,6 +733,9 @@ function computeWorkflowDerived(id, opts) {
     if (!missing.length) {
       steps.push(step('route', 'route', 'done',
         `${lane} lane · specimen test ${sRoute} · surface ${fRoute}`));
+    } else if (effLane !== 'feature') {
+      // The routing questions are asked of feature work only.
+      steps.push(step('route', 'route', 'done', `${lane} lane — routing questions are for features only (${CUT_NOTE})`));
     } else if (legacy) {
       steps.push(step('route', 'route', 'unknown',
         `${lane} lane · ${missing.join(' + ')} not declared (record predates ${ENFORCED_FROM})`));
@@ -875,41 +880,21 @@ function computeWorkflowDerived(id, opts) {
     }
   }
 
-  // ── understand ─ O4: an existing surface is explained BEFORE it is specified ─────────────────
+  // ── understand ─ O4, made optional (sdlc-step-cuts) ───────────────────────────────────────
+  // An explain-surface pass that exists is shown; a missing one never blocks.
   {
     const expl = String(fm.explainer_path || '').trim();
-    if (fRoute !== 'existing') {
-      steps.push(step('understand', 'understand', 'na',
-        fRoute === 'greenfield' ? 'greenfield — nothing existing to explain first' : 'surface check not declared'));
-    } else if (!expl) {
-      steps.push(step('understand', 'understand', legacy ? 'unknown' : 'blocked',
-        'surface_check says existing but no explainer_path is set',
-        'run build-skills/explain-surface.skill in a FRESH context (not the agent that will write the spec), then set explainer_path:',
-        { blocking: !legacy, rule: 'O4' }));
-    } else if (!fs.existsSync(path.join(root, expl))) {
-      steps.push(step('understand', 'understand', legacy ? 'unknown' : 'blocked',
-        `explainer_path '${expl}' does not exist in the repo`,
-        'set explainer_path to a real explain-surface output',
-        { blocking: !legacy, rule: 'O4' }));
+    if (expl && fs.existsSync(path.join(root, expl))) {
+      steps.push(step('understand', 'understand', 'done', `${expl} — optional surface explainer`));
     } else {
-      const eIdx = firstAppearance(root, range, expl);
-      const ok = ordered(eIdx, prdIdx);
-      if (ok === null) {
-        steps.push(step('understand', 'understand', 'unknown',
-          `${expl} present; order undecidable (${range.ok ? 'no spec artifact to order against' : range.reason})`));
-      } else if (ok) {
-        steps.push(step('understand', 'understand', 'done', `${expl} — first-appeared before ${prd}`));
-      } else {
-        steps.push(step('understand', 'understand', legacy ? 'unknown' : 'blocked',
-          `ORDER VIOLATION — the spec (${prd}) first-appeared BEFORE the explainer (${expl})`,
-          'without the pass first, the AGENT acquires the understanding and the PM approves a spec built on facts they never saw',
-          { blocking: !legacy, rule: 'O4' }));
-      }
+      steps.push(step('understand', 'understand', 'na', `the surface explainer is optional — ${CUT_NOTE}`));
     }
   }
 
   // ── spec ─ the lane's artifacts, and O1/O2: they precede the build ───────────────────────────
-  const needsRubric = effLane === 'feature';
+  // A feature PRD's done-criteria section can serve as its rubric (sdlc-step-cuts), so a separate
+  // rubric file is optional in every lane. One that is declared must still exist.
+  const needsRubric = false;
   {
     const wants = {
       feature: 'a PRD', discovery: 'a spike brief', incident: 'an incident record',
@@ -925,7 +910,7 @@ function computeWorkflowDerived(id, opts) {
     } else if (!prd) {
       steps.push(step('spec', 'spec', 'blocked', `no prd_path — the ${effLane} lane needs ${wants}`,
         effLane === 'feature'
-          ? 'author the PRD (skills/core/prd.skill) and the rubric alongside it (D13), then set prd_path + rubric_path'
+          ? 'author the PRD (skills/core/prd.skill) with its done-criteria (D13), then set prd_path'
           : `author ${wants} and set prd_path in ${recRel}`,
         { blocking: true, rule: 'R2' }));
     } else if (prdIdx === null) {
@@ -970,88 +955,24 @@ function computeWorkflowDerived(id, opts) {
   const runs = cachedJsonl(path.join(L.runsDir, L.runsFile)).filter((r) => r && r.initiative_id === id);
   const runsRel = path.relative(root, path.join(L.runsDir, L.runsFile));
 
-  // ── priors ─ O7: the episodic read is CITED, and the citation is checked ─────────────────────
-  // AGENTS.md step 3 mandates reading this initiative's earlier runs BEFORE acting — the one kernel
-  // invariant that was still asserted only in prose. Nothing can derive that a file was READ, and
-  // this rule does not pretend to: what it derives is the three things that ARE decidable from the
-  // record, and together they are what separates it from a checkbox.
-  //   1. whether earlier runs existed at all      — the rule is vacuous on a first run, and says so
-  //   2. whether the run declared what it consulted
-  //   3. whether the ids it named are REAL and actually EARLIER
-  // (3) is the part that matters. O5's `PM_approved: true` is a bare boolean, satisfiable by typing
-  // it; a fabricated `priors` list fails against the corpus it claims to have read. The rule is
-  // NON-BLOCKING for the same reason O5/O6 are: it cannot be true before a run exists, and a rule
-  // enforced at a moment it cannot be satisfied only teaches people to reach for the escape hatch.
-  {
-    const latest = runs.length ? runs[runs.length - 1] : null;
-    const earlier = runs.slice(0, -1);
-    const cite = priorsCitation(latest);
-    const how = `consult this initiative's earlier runs BEFORE acting (AGENTS.md step 3) and cite them as \`priors\` on the run in ${runsRel}`;
-    if (!latest) {
-      steps.push(step('priors', 'priors', 'pending', 'no run logged for this initiative yet', how, { rule: 'O7' }));
-    } else if (!earlier.length) {
-      // Not a pass — there was nothing to pass. A tick here would be the watermelon this engine exists to prevent.
-      steps.push(step('priors', 'priors', 'na',
-        `${latest.run_id} is the first run on this initiative — no priors to consult`));
-    } else if (cite.malformed) {
-      steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-        `${latest.run_id} has a malformed priors citation — ${cite.malformed}`,
-        `use \`priors: ["<run_id>", …]\` or \`priors: {consulted: [...], note: "…"}\` in ${runsRel}`,
-        { rule: 'O7' }));
-    } else if (!cite.declared) {
-      steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-        `${latest.run_id} follows ${earlier.length} earlier run(s) on this initiative and cites no priors`,
-        how, { rule: 'O7' }));
-    } else {
-      const byId = new Map(runs.map((r) => [String(r.run_id), r]));
-      const earlierIds = new Set(earlier.map((r) => String(r.run_id)));
-      const missing = cite.consulted.filter((x) => !byId.has(x));
-      const notEarlier = cite.consulted.filter((x) => byId.has(x) && !earlierIds.has(x));
-      if (missing.length) {
-        // The check a declared field would otherwise never get. A citation naming a run that does not
-        // exist reports a read that cannot have happened — worse than an absent citation, because it
-        // reads as compliance on the board.
-        steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-          `${latest.run_id} cites prior run(s) that do not exist on this initiative: ${missing.join(', ')}`,
-          `cite run_ids that are actually in ${runsRel} — a citation naming nothing reports a read that cannot have happened`,
-          { rule: 'O7' }));
-      } else if (notEarlier.length) {
-        steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-          `${latest.run_id} cites ${notEarlier.join(', ')}, which ${notEarlier.length > 1 ? 'are' : 'is'} not EARLIER than it`,
-          'a run cannot consult itself or a run that came after it — cite only runs logged before this one',
-          { rule: 'O7' }));
-      } else if (!cite.consulted.length && !cite.note) {
-        steps.push(step('priors', 'priors', legacy ? 'pending' : 'blocked',
-          `${latest.run_id} declares an empty priors list with no note, over ${earlier.length} earlier run(s)`,
-          'an empty list with no reason is indistinguishable from not looking — say which priors were read, or say in `note` why none applied',
-          { rule: 'O7' }));
-      } else if (!cite.consulted.length) {
-        // Allowed, and deliberately not silent: the machine cannot judge whether the priors were
-        // genuinely irrelevant, so it surfaces the claim verbatim for the PM instead of ticking.
-        steps.push(step('priors', 'priors', 'done',
-          `${latest.run_id} consulted none of ${earlier.length} earlier run(s) — declared irrelevant: ${cite.note}`));
-      } else {
-        steps.push(step('priors', 'priors', 'done',
-          `${latest.run_id} cites ${cite.consulted.length} prior run(s): ${cite.consulted.join(', ')}${cite.note ? ` — ${cite.note}` : ''}`));
-      }
-    }
-  }
+  // ── priors ─ O7, cut (sdlc-step-cuts) ───────────────────────────────────────────────────────
+  // Reading earlier runs is still good practice; citing them is no longer checked.
+  steps.push(step('priors', 'priors', 'na', `prior-run citation is not required — ${CUT_NOTE}`));
 
-  // ── contract ─ O5: PM-approved done-criteria before the build. Reported, never CI-blocking ────
+  // ── contract ─ O5: PM-approved done-criteria. Reported, never blocking ─────────────────────
+  // Since sdlc-step-cuts only the feature lane asks for it, and a missing sign-off is pending, not
+  // blocked: the PM's approval of the PRD and the acceptance at the end are the sign-offs that count.
   const approved = runs.find((r) => r.pre_run_contract && r.pre_run_contract.PM_approved === true);
   {
     if (approved) {
       steps.push(step('contract', 'contract', 'done',
         `PM-approved pre_run_contract on run ${approved.run_id}${approved.pre_run_contract.approved_by ? ` (by ${approved.pre_run_contract.approved_by})` : ''}`));
-    } else if (runs.length) {
-      steps.push(step('contract', 'contract', legacy ? 'pending' : 'blocked',
-        `${runs.length} run(s) logged, none with pre_run_contract.PM_approved: true`,
-        `agree numbered done_criteria + test_method with the PM and record the approval in ${runsRel}`,
-        { rule: 'O5' }));
+    } else if (effLane !== 'feature') {
+      steps.push(step('contract', 'contract', 'na', `${effLane} lane — the PM's acceptance stands in for a contract sign-off (${CUT_NOTE})`));
     } else {
       steps.push(step('contract', 'contract', 'pending',
-        'no run logged for this initiative yet',
-        `agree done_criteria + test_method BEFORE building and append the run to ${runsRel} with PM_approved: true`,
+        runs.length ? `${runs.length} run(s) logged, none with pre_run_contract.PM_approved: true` : 'no run logged for this initiative yet',
+        `agree done_criteria + test_method with the PM (approving the PRD counts) and record it in ${runsRel}`,
         { rule: 'O5' }));
     }
   }
@@ -1143,41 +1064,32 @@ function computeWorkflowDerived(id, opts) {
        * fix that made a chore's gate visible ADDED one. The module's own rule, repeated in three
        * files — "a skip is an honest RECORD that verification did not happen; it is not
        * verification" — now applies to the surface that reports it. */
+      // The Review Gate runs on request since sdlc-step-cuts: a skip is n/a, and a FAIL is reported
+      // for the PM to weigh at acceptance rather than holding the gate shut.
       if (verdict === 'SKIPPED') {
-        steps.push(step('review', 'review', 'pending', label,
-          'the gate was skipped, so it is still owed — commission an independent evaluator and record a real verdict',
-          { rule: 'R3' }));
+        steps.push(step('review', 'review', 'na', `${label} — skipped; the Review Gate runs on request (${CUT_NOTE})`));
       } else if (verdict === 'FAIL') {
-        steps.push(step('review', 'review', 'blocked', label,
-          'the Review Gate FAILED — remediate the findings and record a new verdict', { rule: 'R3' }));
+        steps.push(step('review', 'review', 'pending', label,
+          'the Review Gate FAILED — weigh the findings at acceptance, or remediate and record a new verdict', { rule: 'R3' }));
       } else {
         steps.push(step('review', 'review', 'done', label));
       }
     } else {
-      steps.push(step('review', 'review', 'pending', 'no Review-Gate verdict recorded',
-        'score the build against the rubric (build-skills/evaluator.skill), then append to evals.jsonl',
-        { rule: 'R3' }));
+      steps.push(step('review', 'review', 'na', `no Review-Gate verdict — it runs on request (${CUT_NOTE})`));
     }
   }
 
-  // ── explain ─ O6: the run is explained AFTER the build and BEFORE acceptance ─────────────────
+  // ── explain ─ O6, made optional (sdlc-step-cuts) ────────────────────────────────────────────
+  // A run explainer that exists is shown; a missing one never blocks acceptance.
   {
     let found = null;
     for (const r of runs) {
       const p = path.join(L.explainers, `${r.run_id}-explainer.md`);
       if (fs.existsSync(p)) { found = path.relative(root, p); break; }
     }
-    if (found) {
-      steps.push(step('explain', 'explain', 'done', found));
-    } else if (!runs.length) {
-      steps.push(step('explain', 'explain', 'pending', 'no run to explain yet',
-        'log the run first, then run build-skills/explain-run.skill against its diff', { rule: 'O6' }));
-    } else {
-      steps.push(step('explain', 'explain', legacy ? 'pending' : 'blocked',
-        `no explainer at ${path.relative(root, L.explainers)}/${runs[runs.length - 1].run_id}-explainer.md`,
-        'run build-skills/explain-run.skill — raw-transcript vigilance degrades within ~30 min, so the explainer is what the PM actually reviews',
-        { rule: 'O6' }));
-    }
+    steps.push(found
+      ? step('explain', 'explain', 'done', found)
+      : step('explain', 'explain', 'na', `the run explainer is optional — ${CUT_NOTE}`));
   }
 
   // ── accept ─ the human PM, final and non-delegable (D12) ─────────────────────────────────────

package/workspace/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "pmos-workspace",
-  "version": "0.78.0",
+  "version": "0.79.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "pmos-workspace",
-      "version": "0.37.0",
+      "version": "0.79.0",
       "license": "MIT",
       "dependencies": {
         "@xterm/addon-fit": "^0.11.0",

package/workspace/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pmos-workspace",
-  "version": "0.78.0",
+  "version": "0.79.0",
   "private": true,
   "description": "PMOS Terminal Workspace — local companion hosting real PTYs beside the library and gates (D62)",
   "license": "MIT",

package/workspace/replaced-tests.json

@@ -123,6 +123,45 @@
         "DOC4: raw shows the file as written, frontmatter fence and all",
         "DOC2: view renders the FILE ON DISK as markdown — a heading, its frontmatter card, its list"
       ]
+    },
+    {
+      "id": "STD-NEXT",
+      "file": "workspace/next-step-actions.test.js",
+      "run": "sdlc-step-cuts",
+      "ref": "DECISIONS.md D93 and planning/initiatives/sdlc-step-cuts.md (PM-approved 2026-10-07)",
+      "reason": "D93 cut the priors step and made the contract optional outside the feature lane, so a chore's reached step is build, not a contract gate. E3, E5, E10 and E11 asserted the cut steps. E2 and E19 keep their intent and now read the build and accept steps of the new ladder; scripts/tests/workflow-order.sh asserts the ladder itself.",
+      "names": [
+        "E3 the reached step is work, and the contract and acceptance are gates",
+        "E5 the sources are the profile-declared ones that suit the step (priors → Confluence)",
+        "E10 the build is refused while the contract is not approved",
+        "E11 the contract is refused as work",
+        "E2 /api/initiative serves an act on every next-step item",
+        "E19 go on a refused step starts nothing"
+      ]
+    },
+    {
+      "id": "STD-PREP",
+      "file": "workspace/proactive-prepare.test.js",
+      "run": "sdlc-step-cuts",
+      "ref": "DECISIONS.md D93 and planning/initiatives/sdlc-step-cuts.md (PM-approved 2026-10-07)",
+      "reason": "D93 made the contract optional outside the feature lane, so a chore no longer waits on a contract gate before build. B7 and B8 asserted that gate. B2, D1 and D6 keep their intent and now name the build step the new ladder reaches.",
+      "names": [
+        "B7 a step the ladder has not reached is refused with the reason (the build waits for the contract)",
+        "B8 a gate step (the contract) can never be queued",
+        "B2 the plan names the agent step the ladder has reached, with the exact prompt, a 16-hex hash and the switch state",
+        "D1 `prepare plan` shows the step, the budget, that the switch is off, the exact prompt and the command to queue it",
+        "D6 `prepare status` lists the entry and the budget"
+      ]
+    },
+    {
+      "id": "STD-TREE",
+      "file": "workspace/first-paint-count.test.js",
+      "run": "sdlc-step-cuts",
+      "ref": "DECISIONS.md D93 and planning/initiatives/sdlc-step-cuts.md (PM-approved 2026-10-07)",
+      "reason": "D93 cut the priors step, so the fixture's finished Spec stage no longer owes a row and folds. The freed space shows the fifth open Build row on first paint: 25 controls against specimen A's 24. The assertion keeps specimen A's count as its base and allows that one row's worth (TREE_PANE_BUDGET = TREE_BUDGET + 1); the (d) setup assertion still pins specimen A at 24.",
+      "names": [
+        "(d) the Tree pane shows at most ${TREE_BUDGET} controls on first paint (specimen A's own Tree view)"
+      ]
     }
   ],
   "entries": [