@ngockhoale/ukit 3.3.2 → 3.4.0

package/CHANGELOG.md

@@ -2,6 +2,48 @@
 
 All notable changes to UKit are documented here.
 
+## 3.4.0 - 2026-09-28
+
+**Playbook system + subagent orchestrator.** Handoff cycles C84–C89 shipped the
+full playbook backlog and the queued orchestrator blueprint:
+
+- **24 playbooks shipped to `template_user/playbooks/`** — bug-fix, investigation,
+  small-feature, feature-implementation (renamed from issue-implementation),
+  refactor, performance, runtime-forensics, session-pickup, verification,
+  skill-evaluation, autonomous-run, architecture-decision, migration, release,
+  prototype + sub-playbooks (hillclimb, babysit, shipping, orchestrate,
+  autopilot-full, autopilot-stack, worktree-cleanup, open-pr) mapped onto
+  existing work-groups via `SUB_PLAYBOOK_WORKGROUP`. `ukit playbook show|list`
+  and `ukit playbook doctor` report all of them; manifest seeds keep installs
+  current.
+- **Routing + measurement**: shared `routeResolver`/`playbookRegistry` twins
+  (byte-parity enforced), session-history extractor, verification-map table +
+  consumers, review-policy decision plane, cross-check depth, fix-loop
+  escalation (`debugLoopThreshold`), measured-reliability tier selection,
+  advisory-turn stop escape (waiting/no-change turns never mint write debt).
+- **`ukit vm run`**: evidence-gated live lane for the Agent VM — real
+  VmEngine+supervisor execution gated behind `decisionRuntime.vm` stages
+  (off→shadow refuse; canary/default required), `lane:'live'` evidence records.
+- **Subagent orchestrator (SO-001..007)**: frozen baseline corpus, `subagent.usage`
+  provenance telemetry (off-default, typed resource provenance, never invented
+  usage), TaskContract/ResultEnvelope validators + bounded artifact refs with
+  claim-level resolution, `probeSubagentCapabilities` host matrix (enforced /
+  advisory / unsupported), typed delegation advice in shadow, context manifest
+  builders (isolated/curated, `fork_unavailable` honest label), golden eval +
+  rollout verdict harness (INCOMPARABLE when evidence < 9-run floor).
+- **Release-cluster contract (RCC-01)**: `docs/pstack/c82/RELEASE_CLUSTER_CONTRACT.md`
+  — `gh`/git-over-Bash fallbacks for every release step + the land-authorization
+  boundary (irreversible writes always ask; Codex gate is advisory).
+- **`worktree-sweep.mjs`**: safety-gated stale-worktree reclamation
+  (enumerate → classify → no-uncommitted/no-active-run gates → remove safe only).
+
+Note: `npm pack` budgets raised once for the wave (packed 3.03MB, unpacked
+9.88MB — attributed in `tests/integration/packageArtifact.test.js`).
+
+## 3.3.3 - 2026-09-27
+
+Republish of 3.3.2 — npm staged-version E409 limbo (same content, version bump only).
+
 ## 3.3.2 - 2026-09-27
 
 **Agent VM driver + self-improve code lane.**  CLI (list/show/status/shadow-run/evidence),  driver producing promotion evidence into , self-improve  step (one shadow run per reference plan per pass), and  step (AUTO-<id>.md handoff task files mined from recurring failure patterns — whitelist fix classes, denylist protected lanes, ≤3/pass, never commits code). See docs/plans/AGENT_VM_DRIVER_PLAN.md + SELF_IMPROVE_CODE_LANE.md.
@@ -13,6 +55,8 @@ All notable changes to UKit are documented here.
 
 - **Self-improve `vmShadow` step (agent-VM driver D-04).** After diagnostics, each rate-limited `runSelfImprove` pass now runs one shadow-run per `planLibrary` reference plan via `agentRuntime/shadowRun.js` when `decisionRuntime.vm.stage` is `'shadow'` or higher — the first live-path caller that accumulates evidence for `PROMOTION_CRITERIA`. `off` → step `skipped` (`stage_off`), zero writes under `.ukit/storage/agent-runtime/`; driver missing, refusing, or throwing → step status `error`, the pass never throws and later steps still run.
 
+- **`ukit vm run` — evidence-gated live lane (BL-031).** `src/cli/commands/vm.js` gains `run <planId>`: a real `VmEngine`+`createSupervisor` path (`runLivePlan` in `src/core/agentRuntime/shadowRun.js`) that launches RUN-node specs declaring the supervisor contract (`spec.command`/`spec.argv` — descriptive `spec.cmd` prose is never shelled out, it fails typed), relays supervised terminal outcomes into the engine, and appends one `lane:'live'` evidence record. Stage gate: `off` → `stage_off` refuse (zero writes); `shadow` → `live_stage_required` refuse — execution requires `canary`/`default`. BL-031 promotion bar is recorded verbatim (`REPLAY_PROMOTION_BAR`: promote only if the VM catches ≥1 defect the doc pipeline missed OR cuts cycle time ≥25% at equal verdicts); `buildReplayComparison`/`recordReplayComparison` write a `replay-comparison` record to the evidence store and `ukit vm evidence` prints its verdict verbatim — `promote()` stays the only stage-advance mechanism.
+
 ## 3.3.1 - 2026-09-27
 
 - **Republish of 3.3.0, content-identical.** npm staged 3.3.0 then returned E409 "previously staged version" on retry — same staged-limbo pattern as 3.0.11→3.0.12. Version bump only; parity verified below.

package/manifests/engineConformance.yaml

@@ -9,8 +9,24 @@
 # Consistency enforced by: tests/consistency/engineConformance.test.js
 version: 1
 routeParity:
-  mechanism: shared-mirror   # all engines route through the same route-task.mjs
+  mechanism: shared-resolver   # one route-task.mjs/resolver serves all engines
   evidence: tests/consistency/engineConformance.test.js
+  # TASK-006 (BL-008): invocation is per-host — the file being shared never
+  # meant Codex auto-routes. Only hook hosts invoke it automatically; Codex's
+  # route is instruction-mediated (advisory), so no 'shared-mirror' auto-route
+  # claim survives here.
+  invocation:
+    claude-code: { mode: hook,        note: 'hooks invoke route-task.mjs automatically on the routed events' }
+    codex:       { mode: instruction, note: 'no hook surface; .codex/README.md instructs running route-task.mjs manually via the .codex/ukit symlink — advisory path' }
+    omp:         { mode: hook,        note: 'omp pre-hook bridge drives the same routing path' }
+# TASK-006 (BL-008): the Codex advisory-fallback contract. On Codex every
+# playbook-covered ask resolves to a measured live path or this advisory
+# fallback — UKit never claims enforcement it lacks.
+codexAdvisoryContract:
+  playbookFile:        'a routed playbookId resolves to the playbook file ~/.ukit/playbooks/<id>.md'
+  outcomeEnforcement:  'outcome records carry enforcement: advisory (emitted by TASK-005 emitters)'
+  ceiling:             'SHADOW/advisory — Codex caps at advisory; never a live-enforcement claim'
+  prohibition:         'no CX cell or surface may claim PROVEN LIVE'
 classes:
   host-enforced:
     description: hook can hard-block (PreToolUse deny, Stop gate, hook deadlines)

package/manifests/hostCapabilities.yaml

@@ -19,7 +19,7 @@ hosts:
     context-files:           { state: tested,    evidence: tests/integration/installPipeline.test.js }
   codex:
     skills:                  { state: observed,  note: '.codex/skills mirror ships via install; no dedicated codex-skill test' }
-    subagents:               { state: unknown }
+    subagents:               { state: observed,  note: 'c84 probe: .codex/agents symlink ships and resolves to the 6 .claude/agents defs; host consumption unverifiable — no codex CLI on host (docs/pstack/c84/codex-capability-probe.md)' }
     hooks:                   { state: unknown }
     session-start-injection: { state: unknown }
     stop-gate:               { state: unknown }
@@ -37,3 +37,70 @@ hosts:
     owner-file-injection:    { state: unknown }
     settings-env:            { state: observed,  note: 'template_project/.omp/config.yml modelRoles reviewed by ompAgentParity' }
     context-files:           { state: observed,  note: 'omp consumes AGENTS.md/CLAUDE.md context layer; no dedicated test' }
+
+# TASK-006 (BL-008): advisory-fallback contract for hosts without a hook
+# surface. Task routing reaches Codex only instruction-mediately — the
+# `.codex/ukit` symlink ships the shared route-task.mjs, but nothing invokes
+# it automatically, so no codex capability cell may rise above what
+# docs/pstack/c84/codex-capability-probe.md measured.
+advisoryContract:
+  ceiling: 'SHADOW/advisory — codex cells stay ≤ observed unless a probe on a host with the codex CLI promotes them; never a live-enforcement claim'
+  outcomeEnforcement: 'codex-engine outcome records carry enforcement: advisory (emitted by TASK-005 emitters)'
+  note: 'codex hook/stop/session-start cells stay unknown; instruction-mediated routing via route-task.mjs is the only path and every outcome record on it carries enforcement: advisory — see manifests/engineConformance.yaml codexAdvisoryContract'
+
+# TASK-C89-004 (SPEC §8 FR-06, §12): subagent-orchestration truth table —
+# what each lane can actually do for a delegated child before lazy
+# disclosure/tool scoping/fork is promised. Same manifest, second family;
+# statuses are the probe vocabulary of probeSubagentCapabilities in
+# src/core/agentRuntime/adapters.js:
+#   supported   — UKit owns the mechanism or cited live evidence proves it
+#   advisory    — a surface exists but UKit cannot bound host consumption
+#   unsupported — the capability does not exist on this lane (no host API)
+#   unverified  — reachable in principle, unproven here (fixture-only or no
+#                 declared live evidence); the only promotable class
+# 'supported' cells require evidenceRef; all others require a note.
+# The manifest records accumulated evidence; a bare runtime probe emits the
+# same cells at default ('unverified' until live evidence is declared).
+subagentOrchestration:
+  probe: 'src/core/agentRuntime/adapters.js probeSubagentCapabilities'
+  lanes:
+    claude-code:
+      spawn:            { status: unverified,  note: 'owned runner spawns via createSupervisor primitives on every lane; live host E2E not declared in-repo — never promoted from primitive presence' }
+      toolFiltering:    { status: advisory,    note: 'agent frontmatter tool lists are host-interpreted hints; UKit cannot bound the child tool surface' }
+      skillCatalog:     { status: advisory,    note: 'skill surface is host-scoped; per-child catalog subsetting is not a host API' }
+      agentCatalog:     { status: advisory,    note: 'host picks a named subagent from the shipped catalog; catalog contents are not subsettable' }
+      writeScope:       { status: advisory,    note: 'artifactDir bounds runner-owned artifact writes; arbitrary child fs writes are not scoped by UKit' }
+      modelBind:        { status: advisory,    note: 'agent frontmatter/modelRoles nominate a model; per-spawn binding is host-discretionary' }
+      fork:             { status: unsupported, note: 'no host state-inheritance API; SPEC §12 fork_unavailable stands' }
+      usageCache:       { status: advisory,    note: 'subagent.usage provenance is observation-only; host provider usage/cache is not readable by UKit' }
+      completionSignal: { status: supported,   evidenceRef: tests/core/agentRuntime/hostAdapters.test.js }
+    codex:
+      spawn:            { status: unverified,  note: 'same owned-runner primitives; live host E2E not declared in-repo' }
+      toolFiltering:    { status: unsupported, note: 'no host tool-filtering API; deny rules are advisory text only (engineConformance codex host-enforced: degraded)' }
+      skillCatalog:     { status: advisory,    note: '.codex/skills mirror ships via install; host consumption unverifiable without codex CLI (docs/pstack/c84/codex-capability-probe.md)' }
+      agentCatalog:     { status: advisory,    note: '.codex/agents symlink resolves to the 6 .claude/agents defs; host consumption INCONCLUSIVE per C84 probe' }
+      writeScope:       { status: unsupported, note: 'no host write-scope API exposed to UKit' }
+      modelBind:        { status: unsupported, note: 'no host model-binding API exposed to UKit' }
+      fork:             { status: unsupported, note: 'no host state-inheritance API; SPEC §12 fork_unavailable stands' }
+      usageCache:       { status: unsupported, note: 'no provider usage/cache surface readable by UKit' }
+      completionSignal: { status: supported,   evidenceRef: tests/core/agentRuntime/hostAdapters.test.js }
+    omp:
+      spawn:            { status: unverified,  note: 'same owned-runner primitives; live host E2E not declared in-repo' }
+      toolFiltering:    { status: advisory,    note: 'omp agent frontmatter is host-interpreted; UKit cannot bound the child tool surface' }
+      skillCatalog:     { status: advisory,    note: 'omp reads .claude/skills via the claude discovery provider; per-child subsetting is not a host API' }
+      agentCatalog:     { status: advisory,    note: 'omp subagent dispatch uses the shipped agent defs; catalog contents are not subsettable' }
+      writeScope:       { status: advisory,    note: 'artifactDir bounds runner-owned artifact writes; arbitrary child fs writes are not scoped by UKit' }
+      modelBind:        { status: advisory,    note: '.omp/config.yml modelRoles nominate tiers; per-spawn binding is host-discretionary' }
+      fork:             { status: unsupported, note: 'no host state-inheritance API; SPEC §12 fork_unavailable stands' }
+      usageCache:       { status: advisory,    note: 'subagent.usage provenance is observation-only; host provider usage/cache is not readable by UKit' }
+      completionSignal: { status: supported,   evidenceRef: tests/core/agentRuntime/hostAdapters.test.js }
+    owned-vm:
+      spawn:            { status: supported,   evidenceRef: tests/core/agentRuntime/shadowRun.test.js, note: 'runLivePlan launches spec.command RUN nodes through the real supervisor; lane gated at decisionRuntime.vm canary/default (src/cli/commands/vm.js)' }
+      toolFiltering:    { status: unsupported, note: 'owned child is a bare subprocess — no tool schema to filter' }
+      skillCatalog:     { status: unsupported, note: 'bare subprocess has no skill catalog to subset' }
+      agentCatalog:     { status: unsupported, note: 'bare subprocess has no agent catalog to subset' }
+      writeScope:       { status: advisory,    note: 'artifactDir bounds runner-owned artifact writes; subprocess fs writes are not sandboxed' }
+      modelBind:        { status: unsupported, note: 'subprocess child has no model surface to bind' }
+      fork:             { status: unsupported, note: 'no state-inheritance surface; supervisor starts a fresh process, never a parent-context fork' }
+      usageCache:       { status: unsupported, note: 'no provider usage/cache surface for a subprocess child' }
+      completionSignal: { status: supported,   evidenceRef: tests/core/agentRuntime/shadowRun.test.js }

package/manifests/platform.full.yaml

@@ -1381,6 +1381,22 @@ items:
     packs:
       - core
 
+  # TASK-003 (FR-003): in-proc SessionStart/session_start chain step that
+  # writes the session's execution.started record. The module itself is a
+  # one-line delegate — the emit implementation ships inside
+  # ukit-runtime-scripts (.claude/ukit/runtime/observability-emit.mjs).
+  - id: hook-observability-emit-module
+    type: hook
+    sourceTemplate: .claude/hooks/observability-emit.mjs
+    targetPath: .claude/hooks/observability-emit.mjs
+    requires:
+      - ukit-runtime-scripts
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
   - id: hook-handoff-model-guard
     type: hook
     sourceTemplate: .claude/hooks/handoff-model-guard.sh
@@ -1399,6 +1415,8 @@ items:
     requires:
       - ukit-index-route-catalog-script
       - ukit-runtime-scripts
+      - ukit-index-route-resolver-script
+      - ukit-index-tier-selection-script
     mergeStrategy: overwrite_with_backup
     variables: []
     enabledByDefault: true
@@ -1597,6 +1615,18 @@ items:
     packs:
       - core
 
+  - id: ukit-index-worktree-sweep-script
+    type: config
+    sourceTemplate: .claude/ukit/index/worktree-sweep.mjs
+    targetPath: .claude/ukit/index/worktree-sweep.mjs
+    requires:
+      - ukit-runtime-scripts
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
   - id: ukit-index-stale-spec-check-script
     type: config
     sourceTemplate: .claude/ukit/index/stale-spec-check.mjs
@@ -1728,6 +1758,109 @@ items:
     packs:
       - core
 
+  - id: ukit-index-route-resolver-script
+    type: config
+    sourceTemplate: .claude/ukit/index/route-resolver.mjs
+    targetPath: .claude/ukit/index/route-resolver.mjs
+    requires:
+      - ukit-index-core-lib
+      - ukit-index-cache-utils-script
+      - ukit-index-route-catalog-script
+      - ukit-index-unic-gateway-script
+      - ukit-index-unic-decision-script
+      - ukit-index-session-history-extractor-script
+      - ukit-index-extract-image-script
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: ukit-index-tier-selection-script
+    type: config
+    sourceTemplate: .claude/ukit/index/tier-selection.mjs
+    targetPath: .claude/ukit/index/tier-selection.mjs
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: ukit-index-session-history-extractor-script
+    type: config
+    sourceTemplate: .claude/ukit/index/session-history-extractor.mjs
+    targetPath: .claude/ukit/index/session-history-extractor.mjs
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: ukit-index-playbook-registry-script
+    type: config
+    sourceTemplate: .claude/ukit/index/playbook-registry.mjs
+    targetPath: .claude/ukit/index/playbook-registry.mjs
+    requires:
+      - ukit-index-route-resolver-script
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: ukit-index-verification-map-script
+    type: config
+    sourceTemplate: .claude/ukit/index/verification-map.mjs
+    targetPath: .claude/ukit/index/verification-map.mjs
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C85 TASK-015 (BL-015): deterministic review decision policy evaluator —
+  # consumed by stop-coordinator.mjs at the pre-Stop gate (advisory verdicts).
+  - id: ukit-index-review-policy-script
+    type: config
+    sourceTemplate: .claude/ukit/index/review-policy.mjs
+    targetPath: .claude/ukit/index/review-policy.mjs
+    requires:
+      - ukit-index-cross-check-matrix-script
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C85 TASK-016 (BL-016): adaptive cross-check depth matrix — resolves
+  # change-risk × implementer-reliability × evidence-quality →
+  # none|runnable|review-round|escalate for the policy's escalationDepth slot.
+  - id: ukit-index-cross-check-matrix-script
+    type: config
+    sourceTemplate: .claude/ukit/index/cross-check-matrix.mjs
+    targetPath: .claude/ukit/index/cross-check-matrix.mjs
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C85 TASK-018 (BL-018): fix-loop escalation to a deeper lane — consumed by
+  # route-task.mjs when the fingerprint-joined fixLoopCount trips
+  # debugLoopThreshold (laneDeepening/verificationDepth hand-off).
+  - id: ukit-index-fix-loop-escalation-script
+    type: config
+    sourceTemplate: .claude/ukit/index/fix-loop-escalation.mjs
+    targetPath: .claude/ukit/index/fix-loop-escalation.mjs
+    requires:
+      - ukit-index-route-resolver-script
+      - ukit-index-cross-check-matrix-script
+    mergeStrategy: overwrite_with_backup
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
   - id: ukit-index-unic-gateway-script
     type: config
     sourceTemplate: .claude/ukit/index/unic-gateway.mjs
@@ -1795,6 +1928,11 @@ items:
       - ukit-index-unic-gateway-script
       - ukit-index-unic-decision-script
       - ukit-index-extract-image-script
+      - ukit-index-route-resolver-script
+      - ukit-index-playbook-registry-script
+      - ukit-index-session-history-extractor-script
+      - ukit-index-fix-loop-escalation-script
+      - ukit-index-tier-selection-script
     mergeStrategy: overwrite_with_backup
     variables: []
     enabledByDefault: true

package/manifests/platform.user.yaml

@@ -52,10 +52,262 @@ items:
     packs:
       - core
 
-  - id: user-playbook-issue
+  - id: user-playbook-feature-implementation
     type: config
-    sourceTemplate: playbooks/issue-implementation.md
-    targetPath: playbooks/issue-implementation.md
+    sourceTemplate: playbooks/feature-implementation.md
+    targetPath: playbooks/feature-implementation.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-investigation
+    type: config
+    sourceTemplate: playbooks/investigation.md
+    targetPath: playbooks/investigation.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-small-feature
+    type: config
+    sourceTemplate: playbooks/small-feature.md
+    targetPath: playbooks/small-feature.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C85 TASK-014 (BL-014): the verification-map table ships beside the
+  # playbooks — JSON, not markdown, so hook consumers query it via the
+  # verification-map.mjs reader instead of parsing prose.
+  - id: user-playbook-verification-map
+    type: config
+    sourceTemplate: playbooks/verification-map.json
+    targetPath: playbooks/verification-map.json
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C86 TASK-020/022 (FR-009): wave-1+3 playbook seeds.
+  - id: user-playbook-performance
+    type: config
+    sourceTemplate: playbooks/performance.md
+    targetPath: playbooks/performance.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-refactor
+    type: config
+    sourceTemplate: playbooks/refactor.md
+    targetPath: playbooks/refactor.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-runtime-forensics
+    type: config
+    sourceTemplate: playbooks/runtime-forensics.md
+    targetPath: playbooks/runtime-forensics.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-session-pickup
+    type: config
+    sourceTemplate: playbooks/session-pickup.md
+    targetPath: playbooks/session-pickup.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-verification
+    type: config
+    sourceTemplate: playbooks/verification.md
+    targetPath: playbooks/verification.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-skill-evaluation
+    type: config
+    sourceTemplate: playbooks/skill-evaluation.md
+    targetPath: playbooks/skill-evaluation.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C87 FR-008: five new playbook seeds (autonomous-run, architecture-decision,
+  # migration, release, prototype) — same config/skip/core shape as the
+  # wave-1 items above.
+  - id: user-playbook-autonomous-run
+    type: config
+    sourceTemplate: playbooks/autonomous-run.md
+    targetPath: playbooks/autonomous-run.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-architecture-decision
+    type: config
+    sourceTemplate: playbooks/architecture-decision.md
+    targetPath: playbooks/architecture-decision.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-migration
+    type: config
+    sourceTemplate: playbooks/migration.md
+    targetPath: playbooks/migration.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-release
+    type: config
+    sourceTemplate: playbooks/release.md
+    targetPath: playbooks/release.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-prototype
+    type: config
+    sourceTemplate: playbooks/prototype.md
+    targetPath: playbooks/prototype.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  # C88 FR-010: eight sub-playbook seeds (hillclimb, babysit, shipping,
+  # orchestrate, autopilot-full, autopilot-stack, worktree-cleanup, open-pr) —
+  # same config/skip/core shape as every user-playbook-* item above.
+  - id: user-playbook-hillclimb
+    type: config
+    sourceTemplate: playbooks/hillclimb.md
+    targetPath: playbooks/hillclimb.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-babysit
+    type: config
+    sourceTemplate: playbooks/babysit.md
+    targetPath: playbooks/babysit.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-shipping
+    type: config
+    sourceTemplate: playbooks/shipping.md
+    targetPath: playbooks/shipping.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-orchestrate
+    type: config
+    sourceTemplate: playbooks/orchestrate.md
+    targetPath: playbooks/orchestrate.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-autopilot-full
+    type: config
+    sourceTemplate: playbooks/autopilot-full.md
+    targetPath: playbooks/autopilot-full.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-autopilot-stack
+    type: config
+    sourceTemplate: playbooks/autopilot-stack.md
+    targetPath: playbooks/autopilot-stack.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-worktree-cleanup
+    type: config
+    sourceTemplate: playbooks/worktree-cleanup.md
+    targetPath: playbooks/worktree-cleanup.md
+    requires: []
+    mergeStrategy: skip
+    variables: []
+    enabledByDefault: true
+    packs:
+      - core
+
+  - id: user-playbook-open-pr
+    type: config
+    sourceTemplate: playbooks/open-pr.md
+    targetPath: playbooks/open-pr.md
     requires: []
     mergeStrategy: skip
     variables: []

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ngockhoale/ukit",
-  "version": "3.3.2",
+  "version": "3.4.0",
   "description": "Install/update an index-first AI workspace for Claude Code, OpenAI Codex and omp (Oh My Pi).",
   "license": "MIT",
   "type": "module",