kiokuko-dsh 0.2.40 → 0.2.42

package/docs/on-demand-intake.md

@@ -1,6 +1,6 @@
 # Conversation-first intake (default)
 
-Conversation-first, on-demand intake is the default. No activation setting is needed for the bundled native host, `createDshHostAdapter`, or `mountCore`; a modular Enno host inherits the core setting. Existing saved configurations that omit `intakeMode` receive this default when loaded. The packaged bundle intentionally needs no new field.
+Conversation-first, on-demand intake is the default. No activation setting is needed for the bundled native host, `createDshHostAdapter`, or `mountCore`; a modular host inherits the core setting. Existing saved configurations that omit `intakeMode` receive this default when loaded. The packaged bundle intentionally needs no new field.
 
 ```ts
 await mountCore(context, {
@@ -9,13 +9,15 @@ await mountCore(context, {
 })
 ```
 
-An explicit `intakeMode: 'eager'` retains upfront preparation for known intent. Native adapters now hand classifier abstention, failure or rejected input to ordinary model reasoning instead of asking the generic task-category question. The original request is retained; explicitly prepared and owned effects still require their preparation and existing host checks. Ordinary native Bash remains on its native permission path, independently of optional preparation. Disabling typed decisions defers interpretation to the main model using the same source-bound contract. Unknown mode values are rejected.
+An explicit `intakeMode: 'eager'` retains upfront preparation for known intent. Native Akinator first asks the configured decision provider to answer each missing intake question. A decision below 51% certainty, abstention, failure or rejected input sends that question to the human through the native question UI. The original request is retained; explicitly prepared and owned effects still require their preparation and existing host checks. Ordinary native Bash remains on its native permission path, independently of optional preparation. Disabling typed decisions defers interpretation to the main model using the same source-bound contract. Unknown mode values are rejected.
 
 A separately supplied custom execution host can use the exported `createDshHostAdapter`, or must implement the same native `onDemandIntake` ownership contract; it cannot be safely synthesized from a tool list. Unsupported custom execution hosts fail with an actionable error rather than silently reverting to upfront questions. They may be updated or deliberately configured with the legacy eager opt-out. Prompt/Skill-only custom hosts own no execution ingress and remain supported.
 
 ## Conversation before execution
 
-The native model receives the complete original user messages and may answer or ask a concrete clarification without opening Akinator's generic purpose question. The initial answer path does not fabricate a chat profile, ready intake, active ledger run, or completed task. Laya's result is provisional intent only; an abstention does not prevent conversation. A queued human request after a normal turn is classified independently rather than inheriting that stale turn's execution readiness. Existing Enno continuations retain their owner and original execution path.
+The native model receives the complete original user messages. Akinator resolves purpose, target and success criteria through decisions first. Answers at or above 51% certainty are accepted; lower certainty, abstention or unavailable decisions ask the human the unresolved question. Probability-based providers use the selected answer's probability; TypeSafe uses its confidence. This threshold applies only to Akinator and does not change execution or permission decisions. Target and success answers must select actual request or current task evidence; the working directory and a generic completion promise are not substitutes. Human and decision answers are labelled separately alongside the unchanged request.
+
+When decisions are explicitly OFF, ordinary conversation proceeds normally, while unresolved action references such as “Fix it” still ask for missing target and success criteria. Prepared task intake likewise leaves missing fields unanswered. Clarification creates no execution authority, active ledger run or completed task. A queued human request after a normal turn is classified independently rather than inheriting that stale turn's execution readiness.
 
 Research, writing and action requests remain work requests. The host guidance tells the assistant to prepare requested tool-backed work rather than silently substitute a text answer or claim unperformed work.
 

package/docs/owned-execution.md

@@ -4,7 +4,7 @@ Kiokuko registers native `kioku_read`, `kioku_write`, `kioku_edit`,
 `kioku_remove`, `kioku_exec`, and `kioku_result` tools. They use Node filesystem
 and process services directly; stock DSH filesystem/shell sandbox settings do
 not control this executor. Native presentation and PTC transport both remain
-supported. Plan, goal, cancellation, admitted identity, and Enno delegation
+supported. Plan, goal, cancellation, admitted identity, and execution owner
 scope still apply.
 
 Ordinary project reads, edits, builds, tests, PATH commands and investigation
@@ -69,12 +69,12 @@ Unrelated generated files and ordinary reads do not invalidate declared source c
 omits it or has `approvalMode:ask`. Ordinary tools and Lisp coexist. Process RPC
 and workspace proposals use the owned executor; helper toolsets still compile
 and validate their schemas/examples. Worker failures leave ordinary tools
-available. `executionMode:protected` explicitly selects the legacy isolated
-worker/approval/fence behavior for compatibility; it is not the default.
-
-Lisp source can directly call filesystem/FFI APIs in development mode. Such
-effects are not independent process/test proof. Prefer `kioku.tools:call-tool`
-and `kioku.process` for observed results. Plan blocks effectful owned/Lisp
+available. Every Lisp worker uses OS isolation: direct filesystem/FFI calls
+cannot access the workspace or inherited credentials. Project effects pass
+through `kioku.tools:call-tool`, `kioku.process` or workspace proposals, where
+the host checks current ownership, permissions and cancellation before acting.
+`executionMode:protected` retains its additional tool fence and proposal policy;
+it is not the default. Plan blocks effectful owned/Lisp
 operations. An explicit human implementation instruction or plan approval
 requests Plan exit at the next accepted native boundary; merely presenting a
 plan does not exit it.

package/docs/pr72-ci-verification.md

@@ -7,8 +7,8 @@ input and `prepare_requested_work`, then retain native permissions, memory
 review, and execution guards.
 
 Native fixtures now prepare each new work request, including human resume
-after unload/reload. The persisted Enno fixture creates the same session
-ownership receipt as production; it does not replace the default with eager
+after unload/reload. The persisted fixtures use the same session
+ownership receipt as production; they do not replace the default with eager
 intake. Eager compatibility tests remain explicitly identified.
 
 The fixed harness retains all original eleven scenario names and assertions.

package/docs/pstack-adaptation.json

@@ -108,20 +108,6 @@
       "changed": "Check commands and examples against current evidence. Separate tested behavior from suggestions. Match requested depth; a plain-language rewrite must not erase important failure modes. Report only the artifact unless explanation is useful.",
       "verification": "tests/dsh/unit/pstack-catalog.test.ts"
     },
-    {
-      "upstream": "pstack/skills/arena/SKILL.md",
-      "destination": "skills/kiokuko-enno-oduno/SKILL.md",
-      "idea": "Compare candidates and partition ownership",
-      "changed": "Host-admitted execution only; no model fallback or automatic fanout",
-      "verification": "tests/dsh/unit/pstack-catalog.test.ts"
-    },
-    {
-      "upstream": "pstack/skills/swarm/SKILL.md",
-      "destination": "skills/kiokuko-enno-oduno/SKILL.md",
-      "idea": "Compare candidates and partition ownership",
-      "changed": "Host-admitted execution only; no model fallback or automatic fanout",
-      "verification": "tests/dsh/unit/pstack-catalog.test.ts"
-    },
     {
       "upstream": "pstack/skills/poteto-mode/SKILL.md",
       "destination": "skills/kiokuko-soul/SKILL.md",
@@ -358,16 +344,6 @@
       "destination": "skills/one-shot-software-completion/SKILL.md",
       "disposition": "Adapt method only; host owns execution, resume and permissions. PR, shipping and cleanup require explicit authorization."
     },
-    {
-      "upstream": "pstack/skills/poteto-mode/playbooks/autopilot-full.md",
-      "destination": "skills/kiokuko-enno-oduno/SKILL.md",
-      "disposition": "Independent coverage only when host admitted; no automatic merging."
-    },
-    {
-      "upstream": "pstack/skills/poteto-mode/playbooks/autopilot-stack.md",
-      "destination": "skills/kiokuko-enno-oduno/SKILL.md",
-      "disposition": "Respect unit dependencies and verify each accepted increment; no inferred Git authority."
-    },
     {
       "upstream": "pstack/skills/poteto-mode/playbooks/babysit.md",
       "destination": "skills/one-shot-software-completion/SKILL.md",
@@ -408,16 +384,6 @@
       "destination": "skills/one-shot-software-completion/SKILL.md",
       "disposition": "Adapt method only; host owns execution, resume and permissions. PR, shipping and cleanup require explicit authorization."
     },
-    {
-      "upstream": "pstack/skills/poteto-mode/playbooks/orchestrate.md",
-      "destination": "skills/kiokuko-enno-oduno/SKILL.md",
-      "disposition": "Only host-admitted WorkUnits; disjoint ownership and evidence-backed synthesis."
-    },
-    {
-      "upstream": "pstack/skills/poteto-mode/playbooks/pause-safely.md",
-      "destination": "skills/kiokuko-enno-oduno/SKILL.md",
-      "disposition": "Preserve completed effects and return through the host pause/recovery contract."
-    },
     {
       "upstream": "pstack/skills/poteto-mode/playbooks/perf-issue.md",
       "destination": "skills/kiokuko-benchmark/SKILL.md",

package/docs/pstack-skills.md

@@ -1,6 +1,6 @@
 # Focused Kiokuko skills
 
-Seven specialists add investigation, architecture, review, benchmarking, verification recipes, skill authoring and technical writing. SOUL selects them only for relevant tasks. Existing code/UI/Enno/memory contracts remain authoritative. See [the adaptation ledger](pstack-adaptation.json) for all 49 upstream skills and 23 playbooks, pinned to the recorded commit. Upstream code is not installed or executed.
+Seven specialists add investigation, architecture, review, benchmarking, verification recipes, skill authoring and technical writing. SOUL selects them only for relevant tasks. Existing code/UI/memory contracts remain authoritative. See [the adaptation ledger](pstack-adaptation.json) for the upstream skills and playbooks, pinned to the recorded commit. Upstream code is not installed or executed.
 
 The common guidance targets GPT-6 Astra and GPT-6.1 Sol without selecting either model or forcing a reasoning budget. Outcomes, constraints and evidence take precedence over ritual steps. Model quality remains **unmeasured** until both actual models are evaluated; local content and delivery tests do not establish model improvement.
 

package/docs/repository-learning.md

@@ -1,22 +1,20 @@
 # Continuous repository learning
 
-Kiokuko automatically checks each **native conversation turn** (`turn/start` through
-`turn/end`), not every token or intermediate model step. Task completion and a
-post-task approval dialog are not prerequisites. Failed, cancelled, no-tool and
-unknown turns are retained as those lifecycle observations, never as verified task
-success.
+Kiokuko offers review after each completed or failed **native conversation turn**, including answers without tools. First it asks whether to organize the turn. It then shows the exact proposed summary, lessons and sources and asks whether to save them. Neither a completed turn nor model output authorizes persistence.
+
+Decline, dismissal, missing question UI, Stop, a new turn and plugin reload save no pending proposal. Cancelled and unknown turns do not open review or create memories. Existing stored memories are preserved.
 
 ## Normal startup and data flow
 
 The public plugin's lifecycle memory mounts the repository-learning coordinator.
 At a turn boundary it freezes bounded, redacted native user/tool/result/assistant
-observations, the native session/turn identity and the current model route. It uses
+observations, the native session/turn identity and the current model route. After the human chooses to organize, it uses
 the host's actual `llm.prepareCall` / `llm.stream` API on that route, with no tools,
 to propose up to four source-cited hypotheses. An externally registered
 `kiokukoReplayReflector` service is **not required** for this path.
 
 The reflector receives the previous repository compact and this turn's evidence.
-It can refine previous hypotheses in light of new evidence. SQLite appends the
+It can refine previous hypotheses in light of new evidence. After approval of the displayed proposal, SQLite appends the
 turn and atomically updates a bounded projection: recent observations and compact
 hypotheses, including provenance, uncertainty, repeated occurrence counts and
 repository tags. A correction may name existing hypothesis IDs in `supersedes`, supply a
@@ -27,8 +25,7 @@ model-proposed corrections, not established truth. Distinct remaining hypotheses
 retain an unresolved-compatibility assessment. History remains in SQLite while the
 model-visible projection has a whole-point budget; omitted points are counted.
 
-Before subsequent model steps, the coordinator waits for preceding repository
-learning and reads that repository's projection directly. **Keyword overlap is not
+A new turn cancels any pending review, then reads the repository's approved projection. **Keyword overlap is not
 required**. A new session in the same canonical repository gets the accumulated
 compact; another repository does not. Root identity comes from trusted host cwd
 and repository resolution, not model-supplied tags. The injected section is
@@ -42,20 +39,17 @@ on the active prompt surface.
 Migration `048_repository_learning.sql` adds the append-only turn ledger and
 bounded repository projection. Idempotent session/turn identities prevent duplicate
 records. Updates use an immediate SQLite transaction and capture-policy revision
-checks. At plugin reload the latest native turn is reconstructed from up to 512
-session events (excluding inherited fork history); already saved turns are not
-reflected again. Earlier unsaved turns beyond this bounded recovery window are not
-claimed recovered.
+checks. At plugin reload, up to 512 session events reconstruct an active turn, excluding inherited history. Historical terminal events do not reopen review or repeat reflection. Approval closures do not survive reload.
 
 Ignorable session plugin records named `plugin:kiokuko-dsh/repository-learning`
 (and live host events `kiokuko/repository-learning`) report capture and retrieval:
 
-- capture: saved, duplicate, excluded or unavailable; revision and reflection status;
+- capture: saved, duplicate, declined, review_unavailable, discarded, excluded or unavailable; revision and reflection status;
 - retrieval: injected, empty or unavailable; injected and omitted point counts.
 
 A missing model route, malformed/ungrounded result or 15-second model deadline
 produces an explicit unavailable/invalid reflection status. The observed turn can
-still be saved without invented model lessons. Database/capture failures are
+still be proposed without invented model lessons, but requires approval before saving. Database/capture failures are
 reported as unavailable, not silently described as successful saving. This feature
 never launches DSH, changes repository source, runs learned commands or grants tool
 permissions.
@@ -85,25 +79,8 @@ process.
 
 ## Interrupted saves and recovery
 
-Reflection has a bounded deadline and no automatic network retry. A disconnected
-or invalid model stream may still save the observed turn with an unavailable or
-invalid reflection, without inventing a lesson. Foreground work can continue.
-SQLite writes retry only recognized lock contention under the existing finite
-backoff policy; no tool or external side effect is replayed by learning recovery.
-
-| Boundary                                               | Durable result and recovery                                                                                                                                                                                         |
-| ------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| Before/during/after the model response                 | Save observations only if reflection fails; the next turn remains usable.                                                                                                                                           |
-| Failure inside the save transaction                    | Roll back turn and projection together. Capture diagnostics report `saveState: not_saved`; reload the plugin to retry the latest retained native turn, or use `/kioku-memory learning reset` to discard it.         |
-| Commit succeeded but acknowledgement was lost          | Re-read the exact repository/session/turn. Report `saved` when found, `unknown` if storage cannot be read. Do not assume rollback or repeat tools. Reload reconciles the stored identity before any new reflection. |
-| Process exits before/after commit                      | SQLite rollback or committed turn survives reopening. The exact identity prevents a second save. Recovery remains limited to the latest 512 retained native events.                                                 |
-| Explicit native Stop or agent/workspace ownership loss | Discard unsaved in-flight reflection, retain only a payload-free retired identity, and reject its late result after reload. Already committed observations are not undone.                                          |
-| Plugin reload                                          | Abort the old coordinator; reconstruct the latest retained turn. A saved turn is a duplicate, an unsaved turn can be retried. No raw reflection response is persisted for retry.                                    |
-| New instruction or next turn                           | Completed earlier turns append in order to the current projection; they do not replace a newer snapshot with an old snapshot. Stop can cancel outstanding reflection so the next instruction can proceed.           |
-| Reset/forget                                           | Existing admission epochs and tombstones reject pre-reset results. Reset also clears current-repository pending admissions; new turns remain enabled.                                                               |
-
-Capture diagnostics retain only safe identities/counts, `saveState`, and a
-`reload_or_reset` recovery hint when a save is absent or uncertain. There is no
-background retry queue or indefinite lease on a model request. A model timeout,
-unknown save outcome or exhausted SQLite retry is not claimed automatically
-recovered. Reset does not erase native logs or user data.
+Reflection has a bounded deadline and no automatic network retry. A failed model call may produce a summary-only proposal; it never permits an automatic save. Foreground work remains usable.
+
+SQLite retries recognized lock contention under its finite backoff policy only while the same approval remains active. An uncertain commit is reconciled against the exact repository/session/turn identity. A failed save is reported as absent or unknown; reload never recreates its approval or retries the old proposal. A new completed turn offers a new review.
+
+Reset, forgetting, native Stop, owner changes and capture exclusions retain their existing fences. Already saved records are not undone by cancelling a later proposal. Native conversation logs are separate from reusable memory and are not erased by declining a save.

package/docs/semantic-compaction.md

@@ -2,7 +2,7 @@
 
 ## Model and reasoning handoff
 
-When a user changes the DSH model or reasoning effort, Kiokuko can shorten eligible older tool results with the selected Jev or Laya decision provider and ask DSH's configured summarizer to create one native checkpoint. The checkpoint replaces older model-visible conversation; the original events remain in the append-only session history. The last two completed model steps, pending input, and latest user message remain verbatim. Automatic Enno/Deep routing and delegated children do not trigger a handoff.
+When a user changes the DSH model or reasoning effort, Kiokuko can shorten eligible older tool results with the selected Jev or Laya decision provider and ask DSH's configured summarizer to create one native checkpoint. The checkpoint replaces older model-visible conversation; the original events remain in the append-only session history. The last two completed model steps, pending input, and latest user message remain verbatim. Automatic Deep routing and delegated children do not trigger a handoff.
 
 ```yaml
 modelHandoff:
@@ -35,7 +35,7 @@ If no complete source line fits that excerpt, the result stays full: no new pack
 
 Failures, unknown success, background execution, Lisp and execution-control results are excluded. Bash additionally requires a captured structured foreground exit status of zero, without interruption or sandbox failure, and an exact match to the logged presentation. `isError: false` alone is insufficient. After reload, Bash results without that live proof remain untouched.
 
-The auxiliary native tool is separate from Enno business operations:
+The auxiliary native tool is separate from business operations:
 
 ```ts
 observation_read({ handle, offset: 2000, limit: 2000 })

package/docs/session-history-repair.md

@@ -41,9 +41,9 @@ hooks do not provide those running services.
 For historical v0 logs, the checker also normalizes Kiokuko's old
 `continuation` / `loop-recovery` message sources and serialized diagnostic
 stacks in supported abort causes. The exact old cancellation text
-`kiokuko dsh Enno continuation stopped: continuation_limit` becomes a
+`kiokuko dsh continuation stopped: continuation_limit` becomes a
 `hook` cause retaining that text; other string causes remain unsupported.
-Native Enno cancellations also use structured `hook` causes for new records.
+Native cancellations also use structured `hook` causes for new records.
 It runs the complete native migration in
 isolation, validates the resulting v3 history, then publishes that new
 generation under DSH's write lease. The original v0 file remains unchanged,

package/docs/skill-compiler.md

@@ -18,9 +18,8 @@ Reload the plugin to change modes. `full` is also the rollback setting. Neither
 mode changes native permissions, tool schemas, task selection or the original
 user message. The ordinary Skill catalog is not a claim that every listed Skill
 was loaded. SOUL is automatic; Japanese guidance follows the routed model family;
-Lisp guidance follows explicit enable; Enno contracts follow the current role and
-selected experts; ordinary additional Skills remain available through the native
-Skill tool. Complete supplied runtime guidance needs no full-file reread.
+Lisp guidance follows explicit enable; ordinary additional Skills remain
+available through the native Skill tool. Complete supplied runtime guidance needs no full-file reread.
 
 ## One source, two representations
 
@@ -77,7 +76,6 @@ it never installs into the user's live profile.
 | --- | --- |
 | SOUL, native/explicit/prompt-only entrypoints | Compiled body in adapter-observed request |
 | All nine native Skill lookups | Native tool execution, result rendering, next request |
-| Enno role Skills and selected expert | Real admitted role transitions, retry and persisted resume |
 | Deep Japanese guidance | Every worker's request and budget accounting |
 | Lisp enable and lisp_describe | Real protected SBCL, six-tool surface, next request body |
 | DeepSeek wire serializer | Actual provider serializer; fetch boundary captured before network |

package/docs/tool-exposure.md

@@ -2,7 +2,7 @@
 
 On hosts where the executor supplies schemas after the prompt waterfall, Kiokuko preserves the provider's admitted tools and reports `unsupported_runtime` for presentation reduction. `phase`, `lean`, and `auto` do not claim tool or description savings on that API. Native permissions, task guards, and scoped protocol tools remain authoritative; no older runtime is selected.
 
-`toolExposure.mode` controls only which Kiokuko-owned definitions appear in native model requests. The default is `auto`. On all DSH native provider connections, normal chat, research, analysis, writing and review omit the nine Kiokuko model tools. Normal build, debug and devops use lean descriptions with the current phase and `nextAction` policy. An active Enno selection takes precedence over task type and uses that same policy. Native and external tools remain available. Configure explicitly with:
+`toolExposure.mode` controls only which Kiokuko-owned definitions appear in native model requests. The default is `auto`. On all DSH native provider connections, normal chat, research, analysis, writing and review omit the Kiokuko model tools entirely. Normal build, debug and devops requests use full definitions without lean descriptions. Native and external tools remain available. Configure explicitly with:
 
 ```yaml
 config:

package/docs/typed-decisions.md

@@ -1,7 +1,7 @@
 # Typed AI decisions
 
 Kiokuko uses one host-owned decision service for provisional Akinator task types,
-installed Skill relevance, past memory reuse, Zenki draft review, [post-display answer review](answer-review.md), and semantic decisions inside Lisp.
+installed Skill relevance, past memory reuse, [post-display answer review](answer-review.md), and semantic decisions inside Lisp.
 The selected adapter owns its transport protocol and uncertainty policy. Domain
 workflows consume `selected` or `abstained`; they do not depend on model names.
 
@@ -145,8 +145,9 @@ shown in the card; selection does not prove authentication or successful inferen
 options as the preset; the native question UI preselects a label ending in
 ` (Recommended)`, so the card still opens editable. A preset grants no authority.
 Immutable existing decision bindings also retain the task choice, without granting
-tool or file authority. The same request and explicit continuation reuse the choice;
-an unrelated new request asks again. Cancel/Skip starts no task. Headless operation with `ask` keeps the
+tool or file authority. Later turns in the same chat and repository reuse the accepted
+choice, including new tasks and session reloads. Cancel/Skip starts no task and is not
+inherited by a later turn. Headless operation with `ask` keeps the
 configured decision mode and explicit manual Lisp preference; `on`/`off` supply that
 answer instead. An explicit
 `typedDecisions.startup: { decision: true, lisp: true }` config selects both and
@@ -286,14 +287,6 @@ missing configuration, uncertainty and service failures use the existing workflo
 fallback. Identity, revision, permission and integrity failures never authorize
 fallback execution.
 
-Zenki calls `enno_plan_review` with the complete candidate. All atomic checks must
-be answered; otherwise the exact persisted `roles.check` model reviews the full
-candidate in a no-tool stream. Both paths share a 30-second review budget. All
-three contribution slots need dispositions. `enno_plan_submit` accepts only the
-identical reviewed candidate with the same catalog and revisions. A changed draft
-requires review again. If neither reviewer completes, the draft remains
-unsubmitted. Existing confirmation and Goki lease checks still apply. Already
-accepted plans continue normally; old pre-draft advice cannot satisfy review.
 
 In Lisp:
 

package/docs/verification.md

@@ -17,14 +17,21 @@ fixtures remain `prerequisite-missing` for checks that require them.
 
 CI schedules **Linux x64** and **macOS ARM64**. Full selection has 21 jobs:
 selection, two native micro-judge builds, 17 verification lanes and the aggregate.
+Every CI run resolves the latest official DSH `master` before selecting coverage.
+That SHA is an independent input to the plan and its digest: repository-only
+diffs, including prose-only changes, cannot reduce GitHub compatibility coverage.
+All lanes build that run's resolved SHA; the next run resolves upstream again.
+Resolution failure fails the run rather than falling back to a locked fixture.
+A daily GitHub run at 03:17 UTC also checks upstream-only changes. Depot keeps its
+Linux-only coverage, but likewise resolves latest DSH and never skips it for prose.
 Linux ARM64 is not scheduled or claimed as CI coverage. Its runtime/build support
 remains in the native target registry; this workflow does not produce its binary.
 
 The required `verify` aggregate consumes `select`, `micro-judge-native` and `checks`.
 Every selected lane and native artifact must be present, successful, fresh and
 bound to the same selected plan and source. Failure, cancellation, a missing
-selected report, an extra target or an altered artifact fails acceptance. Only
-prose-only selection can legitimately skip both executable job groups. The
+selected report, an extra target or an altered artifact fails acceptance. Local
+diff-only plans without an upstream input may skip prose-only work; CI may not. The
 remaining Linux x64/macOS ARM64 check inventories and their deadlines are unchanged.
 The CI review bundle contains exactly its selected targets; standalone native
 bundle verification still requires the complete supported-target inventory.
@@ -82,7 +89,6 @@ arbitrary stdout or a successful no-op process is insufficient.
 | skill-delivery | main, lisp | `node scripts/run-skill-delivery.mjs` | explicit completion receipt |
 | consumer-npm | main, runtime | `node scripts/verify-agenticreplay-consumer.mjs` | explicit completion receipt |
 | consumer-pnpm | main, runtime | `node scripts/verify-agenticreplay-consumer.mjs --pnpm` | explicit completion receipt |
-| dsh-lifecycle | main, runtime | `node scripts/run-enno-dsh-e2e.mjs` | explicit completion receipt |
 | sqlite-vec | main, semantic | `npm run test:sqlite-vec` | explicit completion receipt |
 | code-provider | code-provider | `node scripts/run-ci-native.mjs code-provider` | Explicit selected native fixture; nonempty official lifecycle without skipped required cases. |
 | p0 | p0 | `node scripts/run-ci-native.mjs p0` | Explicit selected native fixture; nonempty official lifecycle without skipped required cases. |
@@ -112,10 +118,6 @@ Node/ABI or native dependency is reused across platforms.
 Import-only convenience entrypoints remain usable explicitly. Recursive discovery
 excludes these aliases and executes each declared leaf exactly once:
 
-- `tests/dsh/integration/capability.test.ts`: `tests/dsh/integration/akinator-gate.test.ts`, `tests/dsh/integration/pre-step/gate.test.ts`, `tests/dsh/unit/capability-catalog.test.ts`.
-- `tests/dsh/integration/close-resume.test.ts`: `tests/dsh/integration/run-close.test.ts`, `tests/dsh/integration/run-resume.test.ts`, `tests/dsh/integration/run-resume/exact-route.test.ts`.
-- `tests/dsh/integration/enno-turn-stopping.test.ts`: `tests/dsh/integration/turn-stopping-gate.test.ts`, `tests/dsh/integration/enno-role-loop.test.ts`.
-- `tests/dsh/integration/provider.test.ts`: `tests/dsh/integration/skill-provider.test.ts`.
 
 ## Explicit optional lanes
 
@@ -143,3 +145,5 @@ Native Lisp integration requires `protected-python3`: a real sandbox subprocess
 probe. An Apple launcher that only reports its version outside the sandbox does
 not satisfy this capability. Local/CI environment preparation must select a real
 installed Python interpreter; the repository never changes the global PATH.
+
+`full` runs unit/integration/e2e through the existing CI ownership split when the protected Lisp lane is required. Lisp and explicitly owned package/Skill cases run in their dedicated checks once. Standalone `main` remains complete. The import-only aliases and manual duplicate suite launchers have been removed; use the canonical check plan.

package/dsh/cordis.legacy.patch.yml

@@ -0,0 +1,7 @@
+# Apply after the Kiokuko bundle only when validating the original plugin mode.
+- id: kiokuko-harness
+  disabled: true
+- id: agent-loop
+  disabled: false
+- id: kiokuko-dsh
+  inject: [skills, systemPrompt, tools, commands, userQuestions, sessions, sessionQuery, sessionPersistence, llm, agents, attachments, subagents, subprocess]

package/dsh/cordis.patch.yml

@@ -3,6 +3,8 @@
 # the stock row automatically.
 - id: session-log-download
   disabled: true
+- id: agent-loop
+  disabled: true
 - insert:
     # Saved model providers must survive a failure of the execution plugin.
     - id: kiokuko-models
@@ -10,7 +12,7 @@
       inject: [llm, commands]
     - id: kiokuko-dsh
       name: 'kiokuko-dsh'
-      inject: [skills, systemPrompt, tools, commands, userQuestions, sessions, sessionQuery, sessionPersistence, llm, agents, attachments, subagents, subprocess]
+      inject: [kiokukoHarness, skills, systemPrompt, tools, commands, userQuestions, sessions, sessionQuery, sessionPersistence, llm, agents, attachments, subagents, subprocess]
       config:
         enabled: true
         agenticReplay:
@@ -21,3 +23,6 @@
           capture:
             content: redacted
             reasoning: false
+    - id: kiokuko-harness
+      name: 'kiokuko-dsh/harness'
+      inject: [agents, sessions, llm, systemPrompt, tools, sessionProjections, sessionPersistence]

package/lisp/supervisor.mjs

@@ -38,12 +38,15 @@ const finish = () => {
   if (!terminal) return
   if (launch.namespaceInfo) {
     if (!namespace) return // Missing ownership evidence never confirms Stop.
-    try {
-      if (sameProcess(namespace.pid, namespace.identity)) {
-        finishTimer ??= setTimeout(() => { finishTimer = undefined; finish() }, 10)
-        return
-      }
-    } catch { unconfirmed('cannot confirm sandbox child termination'); return }
+    let stopped = false
+    try { stopped = !sameProcess(namespace.pid, namespace.identity) }
+    catch (error) { unconfirmed(`cannot confirm sandbox child termination (${error.code ?? 'UNKNOWN'})`) }
+    // Observation can race /proc teardown. Keep ownership until a later positive
+    // confirmation; one failed lookup must not disable the existing polling.
+    if (!stopped) {
+      finishTimer ??= setTimeout(() => { finishTimer = undefined; finish() }, 10)
+      return
+    }
   }
   clearTimeout(finishTimer)
   clearInterval(namespaceLifetime)

package/lisp/tools.lisp

@@ -322,7 +322,7 @@ CHOICES contains 1-8 (restart-name description) pairs. No expression replay or a
         (unless (member restart (compute-restarts condition)) (error "RECOVERY_STALE"))
         (invoke-restart restart)))))
 (in-package :kioku.environment)
-(defun status () (kioku.internal:object "generation" kioku.internal:*generation* "sbcl" (lisp-implementation-version) "os" (software-type) "architecture" (machine-type) "scratch" (namestring (kioku.files:scratch)) "executionMode" (or (uiop:getenv "KIOKU_EXECUTION_MODE") "protected") "network" (if (equal (uiop:getenv "KIOKU_EXECUTION_MODE") "development") "normal user environment" yason:false) "hostWrites" (if (equal (uiop:getenv "KIOKU_EXECUTION_MODE") "development") "owned tools and direct Lisp; only host receipts prove observed effects" "proposals only") "libraries" #("yason" "cl-ppcre" "cl-csv")))
+(defun status () (kioku.internal:object "generation" kioku.internal:*generation* "sbcl" (lisp-implementation-version) "os" (software-type) "architecture" (machine-type) "scratch" (namestring (kioku.files:scratch)) "executionMode" (or (uiop:getenv "KIOKU_EXECUTION_MODE") "protected") "network" yason:false "hostWrites" (if (equal (uiop:getenv "KIOKU_EXECUTION_MODE") "development") "host tools and proposals; operation permissions apply" "proposals only") "libraries" #("yason" "cl-ppcre" "cl-csv")))
 (in-package :kioku.tools)
 (defun available-tools () (kioku.internal:rpc "tools-list" (kioku.internal:object)))
 (defun call-tool (name &optional (arguments (kioku.internal:object)))

package/migrations/051_drop_enno.sql

@@ -0,0 +1,19 @@
+-- 051_drop_enno.sql
+-- The Enno (役小角 / enno-oduno) orchestration feature was removed from the
+-- codebase. Its durable tables are dropped here, children first, so existing
+-- databases stay valid and openable after this forward-only migration.
+-- Shared DSH machinery (ledger_runs, dsh_turn_receipts, dsh_boundary_jobs,
+-- dsh_continuation_outbox, dsh_loop_guard_*) is deliberately retained.
+
+DROP TABLE IF EXISTS enno_advisory_contributions;
+DROP TABLE IF EXISTS enno_advisory_rounds;
+DROP TABLE IF EXISTS enno_execution_leases;
+DROP TABLE IF EXISTS enno_resume_tokens;
+DROP TABLE IF EXISTS enno_dsh_continuations;
+DROP TABLE IF EXISTS enno_operation_receipts;
+DROP TABLE IF EXISTS enno_verifier_runs;
+DROP TABLE IF EXISTS enno_work_units;
+DROP TABLE IF EXISTS enno_plan_drafts;
+DROP TABLE IF EXISTS dsh_enno_delegations;
+DROP TABLE IF EXISTS dsh_enno_memory_refresh;
+DROP TABLE IF EXISTS enno_contracts;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "kiokuko-dsh",
-  "version": "0.2.40",
+  "version": "0.2.42",
   "description": "Kiokuko plugin for DeepSeek Harness",
   "type": "module",
   "license": "MIT",
@@ -22,7 +22,7 @@
     "dist/",
     "migrations/",
     "skills/",
-    "dsh/cordis.patch.yml",
+    "dsh/",
     "docs/",
     "scripts/prepare-build.mjs",
     "scripts/backfill-akinator-memory.mjs",
@@ -70,9 +70,9 @@
       "types": "./dist/dsh/core/index.d.ts",
       "default": "./dist/dsh/core/index.js"
     },
-    "./modules/enno": {
-      "types": "./dist/dsh/modules/enno.d.ts",
-      "default": "./dist/dsh/modules/enno.js"
+    "./harness": {
+      "types": "./dist/dsh/harness/index.d.ts",
+      "default": "./dist/dsh/harness/index.js"
     },
     "./modules/lisp": {
       "types": "./dist/dsh/modules/lisp.d.ts",
@@ -187,14 +187,12 @@
     "test:evaluation:memory-reuse": "node --import tsx scripts/run-memory-reuse-evaluation.mjs",
     "probe:lisp:p0": "node --import tsx scripts/probe-lisp-p0.ts",
     "test:integration": "node scripts/run-tests.mjs tests/dsh/integration",
-    "test:memory:admission": "node scripts/run-tests.mjs tests/dsh/integration/memory-application.test.ts tests/dsh/integration/on-demand-native.test.ts tests/dsh/unit/on-demand-review-regressions.test.ts",
     "test:integration:tool-exposure": "node scripts/run-tests.mjs tests/dsh/integration/tool-exposure-native.test.ts",
     "test:sqlite-vec": "node --import tsx tests/ci/sqlite-vec-package-smoke.ts",
     "test:evaluation": "npm run build && node scripts/run-retrieval-evaluation.mjs",
     "test:benchmark": "npm run build && node scripts/run-retrieval-benchmark.mjs",
     "test:benchmark:dsh-log": "npm run build && node scripts/run-dsh-million-event-benchmark.mjs",
     "test:efficiency": "node --import tsx scripts/run-token-efficiency-evaluation.mjs",
-    "test:e2e:dsh": "npm run build && node scripts/run-enno-dsh-e2e.mjs",
     "test:cli-profile": "node scripts/verify-cli-profile.mjs",
     "verify:prepare": "node scripts/verify-prepare-reproducibility.mjs",
     "sampledb:generate": "node --import tsx tests/fixtures/sample-database.ts tests/sampledb/kiokuko-dsh.sqlite3",
@@ -214,9 +212,6 @@
     "test:benchmark:akinator": "npm run build && node scripts/run-akinator-memory-benchmark.mjs",
     "test:continuity": "node scripts/run-tests.mjs tests/dsh/unit/continuity tests/dsh/integration/continuity",
     "test:evaluation:continuity": "node --import tsx scripts/run-continuity-evaluation.mjs",
-    "test:enno-memory": "node scripts/run-tests.mjs tests/dsh/unit/enno-memory tests/dsh/integration/enno-memory",
-    "test:benchmark:enno-memory": "node --import tsx scripts/run-enno-memory-benchmark.ts",
-    "test:evaluation:enno-memory": "node scripts/run-enno-memory-evaluation.mjs",
     "verify:lisp:vendor": "node scripts/verify-lisp-vendor.mjs",
     "test:packages:live": "KIOKUKO_REQUIRE_LISP_RUNTIME=1 KIOKUKO_PACKAGE_LIVE=1 node scripts/run-tests.mjs tests/dsh/integration/lisp/packages-live.test.ts",
     "test:packages": "KIOKUKO_REQUIRE_LISP_RUNTIME=1 node scripts/run-tests.mjs tests/dsh/unit/lisp/packages.test.ts tests/dsh/integration/lisp/packages-runtime.test.ts tests/dsh/integration/lisp/packages-update.test.ts",
@@ -251,13 +246,9 @@
     "test:ci:e2e": "node scripts/run-ci-native.mjs e2e",
     "test:ci:metadata": "npm run publint",
     "test:ci:e2e:core": "node scripts/run-ci-native.mjs e2e-core",
-    "test:ci:e2e:agent": "node scripts/run-ci-native.mjs e2e-agent",
     "test:ci:e2e:repeated": "node scripts/run-ci-native.mjs e2e-repeated",
     "test:ci:repeated:deep": "node scripts/run-ci-native.mjs e2e-repeated-deep",
     "test:ci:repeated:normal": "node scripts/run-ci-native.mjs e2e-repeated-normal",
-    "test:ci:repeated:enno": "node scripts/run-ci-native.mjs e2e-repeated-enno",
-    "test:ci:repeated:enno:prefix": "node scripts/run-ci-native.mjs e2e-repeated-enno-prefix",
-    "test:ci:repeated:enno:bounded": "node scripts/run-ci-native.mjs e2e-repeated-enno-bounded",
     "test:ci:lisp": "node scripts/run-ci-native.mjs lisp",
     "test:ci:prepare": "npm run verify:prepare",
     "test:ci:embedding": "node scripts/run-local-embedding-smoke.mjs --offline",

package/scripts/session-history-codec.mjs

@@ -156,7 +156,7 @@ function repairLegacyAbortRecord(value) {
   const record = objectRecord(value), data = objectRecord(record?.data)
   const reason = objectRecord(data?.reason)
   if (record?.type !== 'turn/end' || reason?.kind !== 'aborted') return value
-  if (reason.reason === 'kiokuko dsh Enno continuation stopped: continuation_limit') {
+  if (reason.reason === 'kiokuko dsh continuation stopped: continuation_limit') {
     return { ...record, data: { ...data, reason: { ...reason, reason: { kind: 'hook', reason: reason.reason } } } }
   }
   const cause = objectRecord(reason.reason)

package/skills/coding-ideal-routine-skill/SKILL.md

@@ -45,7 +45,7 @@ description: コードを書く、変更するすべての依頼で使用する
 
 ## 2. 部品ではなく、つながった利用経路を見る
 
-DSH の実装タスクで明示的な coding workflow を使う場合は、初回の変更前に `coding_workflow status` を読む。`applicable:false` は通常チャットに workflow がないという観測であり、エラーや通常操作の禁止ではない。準備済みのタスクでは、`propose` で Goal・依存関係付きの全項目・各項目の観測可能な条件・別の最終 Goal 条件を原子的に登録する。先に必要な読取調査を行ってよい。元依頼の権限と対象を広げない。Enno では既存 ideal・plan を提出し、その共通表示を使う。別の Goal を提案しない。
+DSH の実装タスクで明示的な coding workflow を使う場合は、初回の変更前に `coding_workflow status` を読む。`applicable:false` は通常チャットに workflow がないという観測であり、エラーや通常操作の禁止ではない。準備済みのタスクでは、`propose` で Goal・依存関係付きの全項目・各項目の観測可能な条件・別の最終 Goal 条件を原子的に登録する。先に必要な読取調査を行ってよい。元依頼の権限と対象を広げない。既に登録された Goal と検証条件があれば再利用する。
 
 通常 native 実装依頼ではモデル要求前に Lisp を提案し、ON 済みならその経路を使う。Goal 準備は同じ依頼の人間の回答を再利用する。全ての goal.criterionIds を、実装を担当する少なくとも1つの items[].goalCriterionIds に対応付ける。未対応 ID の診断では該当 item を自分で選んで修正し、最終条件を消したり対応を捏造しない。