docs(extension): 收口 setup wizard task packet

tasks/extension-setup-wizard/active/00-state.md

@@ -1,31 +1,28 @@
 # Active State — Runtime-family correction
 
-- **Status:** source implementation authorized for all planned batches
+- **Status:** implementation and preview acceptance complete; merge authorized
 - **Authority:** [D041–D043](../30-decisions-and-questions.md) plus the active
   Impact Handshake and implementation plan
 - **Supersedes:** implementation authority in `79–82`
 - **Evidence:** current inventories in `78`, `80` and Core/Client feature
   worktrees; historical files remain read-only evidence
 
-## Authorized Now
+## Completed Authority
 
-- source implementation for all local batches;
-- investigation and bounded experiments;
-- task-packet maintenance;
-- plan and readiness review.
+- source implementation, cross-repository adoption and preview delivery;
+- commits, pushes and independent package publication;
+- final task-packet maintenance and merge review.
 
-## Not Authorized Yet
+## Remaining Boundary
 
-- commits or pushes;
-- package publication;
-- further cross-repository mutation outside the authorized Core/Client adoption worktrees;
-- preview or production deployment;
-- PR merge.
+- Core PR #65 and Client PR #71 are authorized for squash merge after the final
+  review recorded in the acceptance result.
+- Production Registry publication remains owned by the normal release workflow;
+  this task does not perform a manual production mutation.
 
 ## Current Gate
 
-Batches 1–3 and the Python dependency-acquisition correction are locally
-complete. Core Runtime 0.1.1 must be committed, pushed and published before
-Core PR #65 can receive the released dependency and frozen lock. Its preview
-producer finalization correction is locally ready in the Core worktree. Every
-remote mutation remains separately gated.
+Runtime-family PR #18 is merged. Toolkit 0.2.1, Core Runtime 0.1.1 and Web
+Runtime 0.1.0 are published; Peer locks consume them. Core #65 and Client #71
+are green with successful previews. The current gate is their ordered squash
+merge: Core first, then Client.

tasks/extension-setup-wizard/active/20-batches/runtime-alignment.md

@@ -1,6 +1,6 @@
 # Adjacent Batches — Runtime alignment
 
-- **Status:** planned; each source batch requires its own start authorization
+- **Status:** batches 1–6 complete
 - **Authority:** active Impact Handshake
 - **Supersedes:** the automatic A→G progression implied by plan `81`
 - **Evidence:** active final findings and plan

tasks/extension-setup-wizard/active/40-plan/runtime-alignment.md

@@ -1,9 +1,14 @@
 # Implementation Plan — Runtime alignment
 
-- **Status:** reviewed and ready; source not yet authorized
+- **Status:** implemented; retained as the reviewed pre-implementation plan
 - **Authority:** active Impact Handshake
 - **Supersedes:** plan `81`
-- **Evidence:** active findings, inventories `78`/`80`, accepted D043
+- **Evidence:** active implementation and acceptance results; inventories
+  `78`/`80`; accepted D043
+
+Implementation-result evidence supersedes mechanical details that changed
+during execution, notably the move to one root PDM workspace and the explicitly
+authorized removal of repository test suites.
 
 ## Goal
 
@@ -46,7 +51,7 @@ toolkit/
 src/inkcre_extension_registry/
 ```
 
-Add `runtimes/core-py` to the existing uv workspace and
+Add `runtimes/core-py` to the root PDM workspace and
 `runtimes/client-web` to the pnpm workspace. Root checks build and check all
 four units. The historical `@inkcre/extension-runtime` target/digest API remains
 retired; there is no compatibility package.

tasks/extension-setup-wizard/active/50-implementation/runtime-alignment.md

@@ -1,6 +1,6 @@
 # Implementation Result — Runtime alignment
 
-- **Status:** local implementation complete; released-lock normalization blocked on publication
+- **Status:** implementation, publication and released-lock adoption complete
 - **Authority:** active implementation plan plus the latest authorized batch
 - **Supersedes:** none
 - **Evidence:** repository checks and Peer adoption checks recorded below
@@ -80,7 +80,7 @@
 - All repository, Runtime and Peer tests were explicitly deleted, including
   pytest/Vitest/Playwright dependencies, scripts, fixtures and CI test jobs.
 
-## Remaining gate
+## Release and adoption result
 
 - Runtime-family PR #18 and Web Runtime Version PR #19 were squash-merged.
   Their exact PR heads passed the repository contract.
@@ -93,9 +93,9 @@
   exposed that PDM resolves a relative `--dest` from each selected project.
   Python release destinations now use explicit workspace paths, matching the
   already-correct Web pack destination and the release step's local paths.
-- Publish Toolkit 0.2.0, Python Runtime 0.1.0 and Web Runtime 0.1.0 from the
-  corrected protected-main release run; then update Core PDM and Client pnpm
-  released locks and rerun their frozen full checks.
+- Toolkit 0.2.0/0.2.1, Python Runtime 0.1.0/0.1.1 and Web Runtime 0.1.0 were
+  published from protected main. Core PDM and Client pnpm locks consume the
+  released assets; both repository checks and previews pass.
 
 ## Python dependency acquisition correction
 

tasks/extension-setup-wizard/active/60-acceptance/runtime-alignment.md

@@ -1,18 +1,31 @@
 # Acceptance Result — Runtime alignment
 
-- **Status:** not started
+- **Status:** complete
 - **Authority:** accepted implementation plus the Preview acceptance batch
 - **Supersedes:** none
-- **Evidence:** to be recorded after preview authorization
+- **Evidence:** ext-reg PR #18, Core PR #65, Client PR #71 and their current
+  preview deployments
 
-The acceptance record will contain:
+## Result
 
-1. Core exact installed wheel: disable, make Registry unavailable for the
-   operation, re-enable successfully;
-2. fresh Core environment: local miss resolves and acquires through its
-   configured Registry;
-3. Client Web Runtime: native MF Release loads and the Twitter setup popup
-   opens;
-4. ordinary preview provider success and HTTP non-404 probes only where useful;
-5. defects returned to `active/10-review/` as a new adjacent batch rather than
-   silently expanding the current implementation scope.
+1. ext-reg PR #18 is squash-merged. Toolkit 0.2.1, Core Runtime 0.1.1 and Web
+   Runtime 0.1.0 are published by the managed package workflow.
+2. Core PR #65 passes its repository contract and preview delivery. Its sibling
+   Registry returns `200` for the Twitter 0.2.1 Release and Python Simple index;
+   the deployed Core returns `200` from `/livez` and `/readyz` with runtime
+   phase `ready`.
+3. Client PR #71 passes all repository checks and preview delivery. Its Pages
+   origin returns `200` for the Twitter 0.2.1 Release and MF manifest.
+4. Client #71 connected to Core #65 loads the native Web Distribution and opens
+   the Extension-owned setup popup. Product acceptance reached OAuth and
+   Bookmark Source steps; the final UI corrections retain Extension-owned Close,
+   InkForm fields, direct config projection and operation loading states.
+5. An unavailable exact enabled Distribution during Core cold restore remains
+   observable in logs but no longer takes the base Peer offline or rewrites
+   deployment `enabled[]`. Management remains available for ordinary recovery.
+
+## Verdict
+
+The Extension setup wizard objective and runtime-family correction are complete.
+Final review found no blocking code, contract, documentation or delivery defect.
+Core PR #65 and Client PR #71 are ready for ordered squash merge.

tasks/extension-setup-wizard/packet.md

@@ -30,10 +30,14 @@ Extension config/state and Source/Cron/Job ownership distinct.
   Host Runtime family; Python activation is local-first; `ExtensionManager` and
   `ExtensionBase` belong to that Runtime; Core's rich `ExtensionModel` remains
   the database authority; no Repository or generic Port layer is introduced.
-- Local source implementation and Peer adoption checks are complete. The
-  Runtime-family commit is pushed; package release management is being aligned
-  with Changie/Changesets before PR review. Package publication,
-  released-lock normalization and remote deployment remain separately gated.
+- Runtime-family PR #18 is merged. Toolkit 0.2.1, Core Python Runtime 0.1.1 and
+  Web Runtime 0.1.0 are published through the repository release workflow.
+- Core PR #65 and Client PR #71 consume those releases through their native
+  package locks. Their repository checks and preview deliveries pass; Core is
+  live and ready, and both sibling Registry projections serve Twitter 0.2.1.
+- Final code and documentation review found no blocking defect. Sir authorized
+  squash merge after this review; production Registry publication remains a
+  later release-workflow concern.
 
 ## Current Authority
 
@@ -110,15 +114,11 @@ new chronological design essay:
 6. after authorization, record implementation evidence in `active/50-implementation/`;
 7. record black-box acceptance and remaining findings in `active/60-acceptance/`.
 
-Implementation batches do not automatically authorize the next batch. Commit,
-push, package publication, cross-repository mutation, preview deployment and
-merge remain separate authorization boundaries.
+Implementation batches do not automatically authorize the next batch. The
+completed task retains these boundaries as historical process evidence.
 
 ## Current Gate
 
-Toolkit 0.2.0 and Core Runtime 0.1.0 are Changie baselines; the pending Web
-Runtime changeset will establish 0.1.0 through a Version PR. The next Registry
-gate is reviewing the static Release-route correction, then separately
-authorizing its commit, push and Toolkit 0.2.1 publication before Peer frozen
-locks consume it and rerun preview acceptance. Publication and all
-cross-repository mutations remain separately gated.
+Squash-merge Core PR #65, then Client PR #71. Their managed release workflows
+may subsequently publish the synchronized Twitter 0.2.1 Distributions. No
+manual production Registry mutation belongs to this task.