design-playbook 0.21.1 → 0.22.0

package/README.md

@@ -56,6 +56,9 @@ After install, skills and commands are **namespaced** by the plugin name:
 | `/design-playbook:ux-spec` | Spec-only command |
 | `/design-playbook:ui-review` | Review command |
 | `/design-playbook:run-review` | Cross-run review command |
+| `/design-playbook:run-status` | Run phase, blocker, and resume/continuation narration |
+| `/design-playbook:run-handoff` | Static delivery package for one reviewed run |
+| `/design-playbook:doctor` | Install health diagnosis |
 
 Bare `/design-io` is **not** the installed name — always use the `design-playbook:` prefix.
 
@@ -73,7 +76,7 @@ pi has no plugin namespace — skills are `/skill:<name>`, commands are bare `/<
 | --- | --- |
 | `/skill:design-playbook` | Orchestrator skill (model-invoked) |
 | `/skill:ux-spec` … `/skill:ui-evaluator` | Same eight skills as above |
-| `/design-io` · `/ux-spec` · `/ui-review` · `/run-review` | Pipeline / spec-only / review / cross-run commands |
+| `/design-io` · `/ux-spec` · `/ui-review` · `/run-review` · `/run-status` · `/run-handoff` · `/doctor` | Pipeline / spec-only / review / cross-run / status / handoff / health commands |
 
 pi ships no built-in MCP, so `preview*` and `observe*` skip by default (ADR-0009 absent→skip; the pipeline still runs spec → picker → fill → craft → accept). To enable both gates, install an MCP adapter and register the bundled servers in your project `.mcp.json`:
 
@@ -128,7 +131,7 @@ See the root [README](../../README.md#install-on-other-agents) for the tier tabl
 .mcp.json              ← bundled MCP servers, launched via ${CLAUDE_PLUGIN_ROOT} (ADR-0009)
 mcp/{preview,evidence}/← MCP adapter runtimes (preview_prototype / execute_capture_plan)
 skills/<name>/SKILL.md ← model-invoked skills
-commands/<name>.md     ← slash commands (design-io, ux-spec, ui-review, run-review)
+commands/<name>.md     ← slash commands (design-io, ux-spec, ui-review, run-review, run-status, run-handoff, doctor)
 codex/AGENTS.md        ← Codex bridge notes
 examples/              ← self-authored onboarding samples
 LICENSE · NOTICE       ← authored-only scope
@@ -154,7 +157,15 @@ python <pkg>/scripts/run_status.py .scratch/<run> --json   # machine-readable
 python <pkg>/scripts/run_status.py --list                  # newest runs under .scratch/
 ```
 
-The status command reuses the packaged validator’s G5 confirm rules. It is part of the installed package — not monorepo-only tooling.
+The status command reuses the packaged validator’s G5 confirm rules. It is part of the installed package — not monorepo-only tooling. For an eligible run it also reports an explicit `open-console` continuation command for the local Run Console (it never starts a server itself), with the blocking reason and a safe fallback when the run or Console prerequisites are ineligible. The Console’s current claim is **local, experimental, and trial-gated** — `run-status --json` reports the same capability receipt (`publicClaim: experimental`) — and no authorized external trial or public release is claimed until the separately authorized read-only trial gate passes (ADR-0043).
+
+### Static handoff
+
+```text
+python <pkg>/scripts/run_handoff.py <run> [--fill <declared-path>] [--round N] [--json]
+```
+
+Builds the existing static delivery package for one explicit run from its declared `fill:` path. Multiple declarations need an explicit `--fill`; a missing declaration fails with repair guidance instead of guessing. `verdict` / `authority` / `confirmationSource` are reported verbatim — a `Pending` handoff stays `Pending` and never becomes acceptance.
 
 ### Doctor
 
@@ -163,7 +174,7 @@ python <pkg>/scripts/doctor.py
 python <pkg>/scripts/doctor.py --json
 ```
 
-One packaged diagnosis for interpreter, package surface, optional Playwright, and run-root configuration. Distinguishes `ok` / `degraded` / `broken` with repair actions.
+One packaged diagnosis for interpreter, package surface, optional Playwright, and run-root configuration. Distinguishes `ok` / `degraded` / `broken` with repair actions. Those three states describe **install and runtime health** of what is present locally — they are not a public capability-maturity verdict; maturity vocabulary (`stable` / `experimental` / `blocked-by-gate` / `not-shipped`) stays with the `run-status` capability receipt, and doctor reads existing facts rather than adding a new health or capability-state authority.
 
 **Bundled MCP (v0.3+):** Preview (`mcp/preview/`) and Evidence (`mcp/evidence/`) runtimes ship inside this package and are registered by `.mcp.json` (`${CLAUDE_PLUGIN_ROOT}`). Sibling monorepo dirs remain compatibility launchers/docs. The orchestrator still **probes** MCP `tools/list` and skips `preview*` / `observe*` when tools are absent. Evidence provider writes artifacts only — never the manifest. **`DESIGN_PLAYBOOK_RUN_ROOT`:** default `"."` in `.mcp.json` is the **MCP process cwd**, not the chat workspace — for a host-app dogfood, set an **absolute** path to `.scratch/<run>/` (see [`mcp/evidence/README.md`](mcp/evidence/README.md)). Capture responses include `written_path` (absolute) so mis-rooted writes are visible without a filesystem search.
 

package/codex/AGENTS.md

@@ -1,4 +1,4 @@
-<!-- generated-by design-playbook v0.21.1 -->
+<!-- generated-by design-playbook v0.22.0 -->
 # design-playbook for Codex
 
 ## Install (path of record)
@@ -58,25 +58,23 @@ args = ["<abs>/packages/design-playbook/mcp/evidence/server.py"]
 Verify: `codex mcp list` should list both. `preview*` needs a system Edge/Chrome (the
 adapter spawns it via `--app=`); `observe*` needs Playwright + Chromium.
 
-> **`preview*` silently skips when `preview_prototype` is absent.** If preview does not
-> appear, the orchestrator probed `tools/list`, found no `preview_prototype`, and skipped
-> G5 - this is designed skip behaviour, not a crash. Confirm the tool is registered
+> **`preview*` skips when `preview_prototype` is absent; G5 is not triggered.**
+> Follow the orchestrator's skip narration to report the reason and enable path.
+> This is designed skip behaviour, not a crash. Confirm the tool is registered
 > (`codex mcp list`) before treating it as a preview failure. Codex end-to-end preview
 > smoke is not yet validated (v0.4.4 deferred the codex E2E smoke; only evidence/G6 was
 > server-level smoked).
 
 ## Load order
 
-1. `skills/design-playbook/SKILL.md`
-2. Standard order: `design-baseline?` → `ux-spec` → `ui-picker` → `fill` → `craft-guard` → `ui-evaluator`.
+Read `skills/design-playbook/SKILL.md` first. Its **Run profile** and **Steps**
+sections are the sole authority for stage order, conditional entry, and completion
+criteria. Consume the `run_profile.py route` result as directed there; this bridge
+does not define a second route or a host-specific sequence.
 
-Native desktop order: `ux-spec` → `native-craft` → `ui-picker` → `fill` → `craft-guard` → `ui-evaluator`.
-
-Conditional entry `design-baseline?` (ADR-0012) runs before `reference-intake?` when the router returns `requires_baseline`. Existing-product Fill requires a valid existing baseline, an accepted generated baseline, or an explicit user waiver.
-
-Conditional entry `reference-intake?` (screenshot/URL/design/product analogy, ADR-0011) runs **before** `ux-spec?` when the router returns `requires_reference_contract` — fixed orchestrator order, not reorderable. Run `native-craft` only for an explicit native-desktop/native-feel target. Web and mobile Web skip `native-craft`; if the platform is unclear, ask before choosing the order. The orchestrator owns the decision gate, render-surface seam handoff, and fail-closed behavior.
-
-Mirror the orchestrator's skip narration (SKILL.md Steps preamble): when a step is skipped, output one line — step name + reason + how to enable, with the gate label when one applies, e.g. `-> preview*: adapter absent, skipped (G5 not triggered; enable via packages/design-playbook/mcp/preview/ or host MCP)`.
+For optional adapters, follow the orchestrator's **Steps** preamble and
+`skills/design-playbook/references/load-map.md` pointers. Use its one-line skip
+narration when a stage does not run, including the reason and enable path.
 
 Audit preferences (ADR-0033) apply identically on Codex. Follow `skills/design-playbook/SKILL.md` § *Audit preferences* as sole authority; this bridge adds no host-specific preference rules.
 

package/commands/doctor.md

@@ -14,3 +14,5 @@ python <plugin>/scripts/doctor.py --repo-root <target-repo>
 ```
 
 Reports `ok` / `degraded` / `broken`. Failed checks include a concrete repair. Optional adapters (Playwright, run-root env) degrade rather than hard-fail the install. Audit-preference state shows effective stage values, sources, asked status, and corrupt layers for target repository.
+
+`ok` / `degraded` / `broken` describe install and runtime health of the local surface only — they are not a public capability-maturity verdict. Maturity vocabulary (`stable` / `experimental` / `blocked-by-gate` / `not-shipped`) stays with the `run-status` capability receipt; doctor reads existing facts and adds no new health or capability-state authority.

package/commands/run-handoff.md

@@ -0,0 +1,27 @@
+---
+description: Build a static run handoff from an explicit run and its declared Fill
+---
+
+# run-handoff
+
+Build the existing static delivery package for one explicit Design I/O run. Not a pipeline obligation and not acceptance.
+
+## Usage
+
+```text
+python <plugin>/scripts/run_handoff.py <run> [--fill <declared-path>] [--round N] [--summary <text>] [--lang zh-CN|en] [--json]
+```
+
+- `<run>` is required. Do not discover a run from `.scratch/`.
+- Read `fill:` declarations from that run's `plan.md` (unfenced column-0 lines). One declared Fill is used automatically; multiple require `--fill` with one of those declared paths; none or a missing/ineligible declaration fails with repair guidance.
+- Do not scan the project for HTML, and do not use preview or reference assets as the reviewed Fill.
+- The existing Evidence builder writes `evidence/static-handoff/` (delivery page, Fill copy, archive, disclosure, snapshots). Report `verdict`, `authority`, and `confirmationSource` verbatim, including `Pending`, `blocked`, `not-applicable`, and `unsubstantiated`.
+- Do not write acceptance or convert a non-Pass state into Pass.
+- Page language: explicit `--lang`, then `DPB_PREVIEW_LANG`, then `LANG`, otherwise Chinese. Unknown environment locales use the same Chinese default as Preview. Language affects UI labels only; machine fields, original diagnostics, and user-authored content remain verbatim.
+
+## Done when
+
+The user asked for a handoff; `index.html` exists under the selected run tree; and `verdict` / `authority` / `confirmationSource` were reported verbatim.
+
+Scope:
+$ARGUMENTS

package/commands/run-status.md

@@ -13,9 +13,11 @@ python <plugin>/scripts/run_status.py [.scratch/<run>] [--json] [--list] [--scra
 ```
 
 - Omit the run path to pick the newest directory under `--scratch` (default `.scratch/`).
-- `--json` emits machine-readable stages + `next` action.
+- `--json` emits machine-readable stages + `next` action, plus the continuation block: selected run, current phase, blocker, run integrity, and the Console capability receipt (`implementation` / `validation` / `availability` / `publicClaim`).
 - `--list` only lists discovered runs.
 
+For an eligible run the continuation names an explicit `open-console` command for the local, experimental, trial-gated Run Console — no authorized external trial or public release is claimed until the separately authorized read-only trial gate passes (ADR-0043). `run-status` never starts a server, daemon, or background process itself; an ineligible run reports the blocking reason and a safe fallback instead.
+
 ## Done when
 
-The command names completed stage markers, any active blocker (preview floor, baseline gate, recirculate verdict), and the single next valid resume action. It reuses `validate_run` judgments for G5 confirm validity rather than inventing a second state machine.
+The command names completed stage markers, any active blocker (preview floor, baseline gate, recirculate verdict), and the single next valid resume action. It reuses `validate_run` judgments for G5 confirm validity rather than inventing a second state machine. Stale, partial, hash-mismatched, malformed, or inconsistent runs stay visible as those states and are never replaced by an older successful snapshot.

package/commands/ui-review.md

@@ -2,7 +2,7 @@
 description: Dual-track UI review emitting the six-block point-back report (ledger / findings / positives / coverage / limitations / verdict)
 ---
 
-Run skill **ui-evaluator** (pull craft-guard checks when AI slop/motion/loading is in scope). Output issue/source/fix/severity; blocking first.
+Run skill **ui-evaluator** (pull craft-guard checks when AI slop/motion/loading is in scope). Output issue/source/fix/severity/track; blocking first.
 
 Scope:
 $ARGUMENTS

package/examples/dogfood/run/craft-guard.md

@@ -17,3 +17,12 @@ Registry: `skills/design-playbook/references/rules.md`, full catalog (P3 run:
 | I18N-01@1 | not-applicable | 单语控制台（zh-CN），无 i18n 声明（无 i18n.* 契约字段，L1 未声明多语言用户群） | - | - | - | 单语声明成立 | - |
 | PERF-01@1 | applicable | - | clear | 长运行有持续进度感（feed 条目级进度逐 tick 更新；全局暂停 busy 即时反馈） | evidence/L6.2-pause-trace.json 反馈序列 | 反馈相称性未承诺耗时阈值（契约无阈值声明） | - |
 | SEC-01@1 | not-applicable | 声明范围无敏感操作新增（敏感模拟参数默认脱敏沿用；全局暂停非敏感操作） | - | - | - | 无敏感面可查 | - |
+| COPY-01@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档（2026-08-14）；主动语态与动作命名一致性审查所需的全流程文案清单未采集 | - | - | - | - | - |
+| COPY-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；用户侧命名审查所需的界面名词与实现命名对照未采集 | - | - | - | - | - |
+| COPY-03@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；错误信息语气审查所需的错误态文案样本未采集 | - | - | - | - | - |
+| A11Y-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；可见键盘焦点判定所需的聚焦态截图与键盘走查未采集（a11y 树无法证明视觉属性） | - | - | - | - | - |
+| CRAFT-09@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；归档缺审计面样式源——filled-ui.md 为声明性产物索引不含样式源，candidates/preview 为一次性原型资产而非填充面源码 | - | - | - | - | - |
+| CRAFT-10@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；归档缺审计面源标记与结构装置的视觉捕捉——evidence 仅交互轨迹与 a11y 树 JSON，无填充面标记源归档 | - | - | - | - | - |
+| DECIDE-01@1 | applicable | - | clear | 选中方向为全局 run console 构成重组（candidates/console-region.html 草图 + preview round 1/2 用户确认），非未审视的默认外观收敛 | DD-0001 理由可回溯 l1.scenes（切页续读）与 PERF-01 比较轴及经用户确认的布局段突破；DD-0002 理由可回溯 l6.c4（跨视图状态闭环）与运行中心第一步方向——均引用 brief 具体事实 | 常规方向经比较矩阵沿 brief 轴证成，非未审视默认；无基线默认方向身份声明 | - |
+
+注：2026-08-28 注册批（COPY-01/02/03、A11Y-02、CRAFT-09/10、DECIDE-01）晚于本 run 存档；按三态谓词补记——blocked 行在理由列点名缺失的证据面；DECIDE-01 依归档内可读的决策报告求值为 applicable。

package/examples/export-entry/run/craft-guard.md

@@ -17,3 +17,12 @@ Registry: `skills/design-playbook/references/rules.md`, full catalog (P2 run). S
 | I18N-01@1 | not-applicable | 单语控制台，无 i18n 声明（无 i18n.* 契约字段，L1 未声明多语言用户群） | - | - | - | 单语声明成立 | - |
 | PERF-01@1 | blocked | 性能感知需运行时度量，本 run provider 缺度量面（measurement 层不可采） | - | 导出等待仅观察到 busy 态 | 度量面缺席，无法判定反馈与耗时的相称性 | 无法在不承诺阈值的情况下检查例外 | 补采运行时度量后重评；缺口的证据语义见 point-back 覆盖声明 |
 | SEC-01@1 | not-applicable | 声明范围无敏感操作新增（导出非敏感数据；隐藏敏感列由 column_scope 假设排除） | - | - | - | 无敏感面可查 | - |
+| COPY-01@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档（2026-08-14）；主动语态与动作命名一致性审查所需的全流程文案清单未采集 | - | - | - | - | - |
+| COPY-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；用户侧命名审查所需的界面名词与实现命名对照未采集 | - | - | - | - | - |
+| COPY-03@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；错误信息语气审查所需的错误态文案样本未采集 | - | - | - | - | - |
+| A11Y-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；可见键盘焦点判定所需的聚焦态截图与键盘走查未采集（a11y 树无法证明视觉属性） | - | - | - | - | - |
+| CRAFT-09@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；归档缺审计面样式源——filled-ui.md 为静态替身描述，无填充面样式源文件归档 | - | - | - | - | - |
+| CRAFT-10@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；归档缺审计面源标记——filled-ui.md 为静态替身描述；既有截图仅覆盖错误态，无结构装置的视觉捕捉 | - | - | - | - | - |
+| DECIDE-01@1 | applicable | - | clear | 选中方向为按周命名模式（compare 档轻量比较），不落入 2026-08 记录的自默认外观样貌；DD-0001 为 record 档，不在本规则范围 | DD-0002 理由可回溯 l1.target_user 比较轴（周频归档场景下检索是主任务）——引用 brief 具体事实而非通用措辞 | 常规方向经比较矩阵沿 brief 轴证成，非未审视默认；无基线默认方向身份声明 | - |
+
+注：2026-08-28 注册批（COPY-01/02/03、A11Y-02、CRAFT-09/10、DECIDE-01）晚于本 run 存档；按三态谓词补记——blocked 行在理由列点名缺失的证据面；DECIDE-01 依归档内可读的决策报告求值为 applicable。

package/examples/export-pointfix-upgrade/run/craft-guard.md

@@ -17,3 +17,12 @@ Registry: `skills/design-playbook/references/rules.md`. Seven-column rows; after
 | I18N-01@1 | not-applicable | 单语控制台，无 i18n 声明（无 i18n.* 契约字段，L1 未声明多语言用户群） | - | - | - | 单语声明成立 | - |
 | PERF-01@1 | blocked | 性能感知需运行时度量，本 run provider 缺度量面（measurement 层不可采） | - | 导出等待仅观察到 busy 态 | 度量面缺席，无法判定反馈与耗时的相称性 | 无法在不承诺阈值的情况下检查例外 | 补采运行时度量后重评；缺口的证据语义见 point-back 覆盖声明 |
 | SEC-01@1 | not-applicable | 声明范围无敏感操作新增（导出非敏感数据；隐藏敏感列由 column_scope 假设排除） | - | - | - | 无敏感面可查 | - |
+| COPY-01@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档（2026-08-14）；主动语态与动作命名一致性审查所需的全流程文案清单未采集 | - | - | - | - | - |
+| COPY-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；用户侧命名审查所需的界面名词与实现命名对照未采集 | - | - | - | - | - |
+| COPY-03@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；错误信息语气审查所需的错误态文案样本未采集 | - | - | - | - | - |
+| A11Y-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；可见键盘焦点判定所需的聚焦态截图与键盘走查未采集（a11y 树无法证明视觉属性） | - | - | - | - | - |
+| CRAFT-09@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；选择器优先级冲突审查所需的样式源走查未执行 | - | - | - | - | - |
+| CRAFT-10@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；结构装置与内容属性对应关系的走查未执行 | - | - | - | - | - |
+| DECIDE-01@1 | not-applicable | 决策报告仅含 record 档 DD-0101，无 compare/explore 档方向决策条目 | - | - | - | - | - |
+
+注：2026-08-28 注册批（COPY-01/02/03、A11Y-02、CRAFT-09/10、DECIDE-01）晚于本 run 存档；按三态谓词补记，blocked 行在理由列记缺失证据面。

package/examples/export-upgrade/run/craft-guard.md

@@ -17,3 +17,12 @@ Registry: `skills/design-playbook/references/rules.md`, full catalog (P3 run:
 | I18N-01@1 | not-applicable | 单语控制台，无 i18n 声明（无 i18n.* 契约字段，L1 未声明多语言用户群） | - | - | - | 单语声明成立 | - |
 | PERF-01@1 | applicable | - | clear | 长导出有持续进度感（30s 窗口 5 次采样，条目级进度持续更新） | evidence/L6.1-status-trace.json 进度采样序列 | 反馈相称性未承诺耗时阈值（契约无阈值声明） | - |
 | SEC-01@1 | not-applicable | 声明范围无敏感操作新增（导出非敏感数据；隐藏敏感列由 column_scope 假设排除） | - | - | - | 无敏感面可查 | - |
+| COPY-01@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档（2026-08-14）；主动语态与动作命名一致性审查所需的全流程文案清单未采集 | - | - | - | - | - |
+| COPY-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；用户侧命名审查所需的界面名词与实现命名对照未采集 | - | - | - | - | - |
+| COPY-03@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；错误信息语气审查所需的错误态文案样本未采集 | - | - | - | - | - |
+| A11Y-02@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；可见键盘焦点判定所需的聚焦态截图与键盘走查未采集（a11y 树无法证明视觉属性） | - | - | - | - | - |
+| CRAFT-09@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；归档缺审计面样式源——无填充产物源文件归档，candidates/preview 为一次性原型资产而非填充面源码 | - | - | - | - | - |
+| CRAFT-10@1 | blocked | 条目 2026-08-28 注册，晚于本 run 存档；归档缺审计面源标记与结构装置的视觉捕捉——evidence 仅交互轨迹 JSON，无填充产物源文件归档 | - | - | - | - | - |
+| DECIDE-01@1 | applicable | - | clear | 选中方向为启用既有 status region 收纳导出任务（candidates/B.html 草图 + preview round 1/2 用户确认），非未审视的默认外观收敛 | DD-0003 理由可回溯 l1.scenes（导出中切页全局可查）与 PERF-01 比较轴；DD-0004 理由可回溯 l6.c2（跨视图状态闭环）与基线 status region 惯例声明——均引用 brief 具体事实 | 常规方向经比较矩阵沿 brief 轴证成，非未审视默认；基线声明的是 status region 惯例而非默认外观身份 | - |
+
+注：2026-08-28 注册批（COPY-01/02/03、A11Y-02、CRAFT-09/10、DECIDE-01）晚于本 run 存档；按三态谓词补记——blocked 行在理由列点名缺失的证据面；DECIDE-01 依归档内可读的决策报告求值为 applicable。

package/lib/index.js

@@ -8,11 +8,11 @@
  *    `skills/` directory. The plugin locates the directory via `__dirname`,
  *    so no `!!js` expression and no cwd-dependent resolution is involved.
  *
- * 2. Six slash commands (ctx.commands) — `design-io`, `doctor`,
- *    `run-review`, `run-status`, `ui-review`, `ux-spec` — that load the
- *    matching `commands/<name>.md` prompt, substitute `$ARGUMENTS` with the
- *    raw trailing input, and inject it as a user-role follow-up turn via
- *    `agent.followup()`.
+ * 2. Slash commands (ctx.commands) — `design-io`, `doctor`,
+ *    `run-handoff`, `run-review`, `run-status`, `ui-review`, `ux-spec` —
+ *    that load the matching `commands/<name>.md` prompt, substitute
+ *    `$ARGUMENTS` with the raw trailing input, and inject it as a
+ *    user-role follow-up turn via `agent.followup()`.
  *
  * The Cordis `!!js` evaluation scope provides no `require` (only Node globals
  * plus ctx-provided values like dshHomePath/loader), so pointing a
@@ -81,7 +81,7 @@ exports.parseSkillFile = parseSkillFile
 exports.createUserMessageFromPrompt = createUserMessageFromPrompt
 
 /**
- * The six slash commands this plugin registers. Each maps to a
+ * Slash commands this plugin registers. Each maps to a
  * `commands/<name>.md` prompt file; the file's frontmatter `description`
  * becomes the command's discovery metadata, and the body (with `$ARGUMENTS`
  * substituted) is injected as a user follow-up turn.
@@ -89,6 +89,7 @@ exports.createUserMessageFromPrompt = createUserMessageFromPrompt
 const COMMAND_NAMES = [
   'design-io',
   'doctor',
+  'run-handoff',
   'run-review',
   'run-status',
   'ui-review',

package/mcp/evidence/capture_runtime.py

@@ -662,7 +662,7 @@ def matrix_viewport(name: str) -> dict[str, Any]:
 def capture_delivery_matrix(
     *,
     url: str,
-    out_dir: Path | None = None,
+    out_dir: Path,
     freeze: dict[str, Any] | None = None,
     browser_adapter: BrowserAdapter | None = None,
 ) -> dict[str, dict[str, Any]]:
@@ -674,19 +674,19 @@ def capture_delivery_matrix(
     can assemble the ``disclosure-review.json`` matrix and the ``/export-zip``
     package from one pass.
 
-    ``out_dir`` defaults to the spec's canonical delivery path
-    ``output/playwright/static-handoff/`` (spec §5). ``browser_adapter`` uses
-    the same ``BrowserAdapter`` seam as ``execute_capture_plan`` (real
-    Playwright by default, injected fake in tests). Production adapters should
-    expose ``capture_and_probe`` so the snapshot and metrics share one page;
-    adapters without that seam remain explicitly ``unmeasured``.
+    ``out_dir`` is required and carries no default: handoff artifacts live
+    under the run tree (``<run_root>/evidence/static-handoff/``, ADR-0034 §5),
+    never the process CWD, so the caller must pass the run-tree destination
+    explicitly. ``browser_adapter`` uses the same ``BrowserAdapter`` seam as
+    ``execute_capture_plan`` (real Playwright by default, injected fake in
+    tests). Production adapters should expose ``capture_and_probe`` so the
+    snapshot and metrics share one page; adapters without that seam remain
+    explicitly ``unmeasured``.
     """
     freeze = freeze or {"enabled": True, "waitFonts": True, "networkIdle": False}
     if browser_adapter is None:
         browser_adapter = PlaywrightBrowserAdapter()
 
-    if out_dir is None:
-        out_dir = Path("output/playwright/static-handoff")
     out_dir = Path(out_dir)
     out_dir.mkdir(parents=True, exist_ok=True)
     results: dict[str, dict[str, Any]] = {}

package/mcp/evidence/disclosure.py

@@ -19,8 +19,10 @@ Two seams keep the contract builder pure and testable without a browser:
   It only normalizes caller-supplied facts into the §4.2 shape.
 
 ``build_handoff_zip()`` packages the disclosure credential plus any caller-
-supplied snapshot artifacts into a single ZIP for the local ``/export-zip``
-endpoint (Stage 9 delivery mount). It never reads outside the caller-provided
+supplied snapshot artifacts into a single ZIP. The Evidence-side builder
+(``handoff.py``) writes it to disk as ``static-handoff.zip`` next to the
+delivery page under ``<run_root>/evidence/static-handoff/`` (ADR-0034); no
+HTTP delivery endpoint exists. It never reads outside the caller-provided
 file list.
 """
 

package/mcp/evidence/handoff.py

@@ -10,6 +10,12 @@ ADR-0034 fixed three boundaries the first Stage 9 implementation crossed:
   ``confirm-round-*.json`` that ``transaction.py`` persists (ADR-0013), never
   re-derived. The ADR-0008 floor lives inside that record; re-deriving it here
   would be a second confirmation authority, which CONTEXT.md forbids.
+- **Verdict authority** - ``point-back.md`` owns the verdict (continuation
+  pack decision 4): a delivery Pass additionally requires the point-back's
+  own canonical Pass verdict, read through the single verdict-syntax parser
+  (ADR-0025) shared with G3 and run status. A confirmed round with resolved
+  gates can never convert a Recirculate/blocked - or missing/ambiguous -
+  owner verdict into Pass (decision 25).
 - **Capture target** - the five-viewport matrix and the layout probe run
   against the deliverable itself, not against any review chrome.
 
@@ -33,6 +39,8 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import Any, Callable
 
+from design_playbook.mcp.ui_locale import resolve_ui_locale
+from design_playbook.mcp.evidence.handoff_i18n import STRINGS
 from design_playbook.mcp.evidence.disclosure import (
     VIEWPORT_ORDER,
     ViewportMetrics,
@@ -59,6 +67,7 @@ class StaticHandoffResult:
     json_path: Path
     zip_path: Path
     index_html: Path
+    deliverable_html: Path
 
 
 def _iso_now() -> str:
@@ -117,6 +126,25 @@ def _confirmation_from_record(
     return True, ""
 
 
+def _point_back_verdict(run_root: Path) -> str | None:
+    """The canonical point-back verdict, or ``None`` when there is not
+    exactly one.
+
+    point-back.md is the finding/verdict authority (continuation pack
+    decision 4). Reading goes through the single verdict-syntax parser
+    (ADR-0025) that G3 and run status already share - never a second
+    parser here. ``None`` (missing, malformed, ambiguous, or repeated
+    verdict text) is never a Pass.
+    """
+    from design_playbook.scripts.verdict_syntax import parse_verdict
+
+    try:
+        text = (run_root / "point-back.md").read_text(encoding="utf-8")
+    except (OSError, UnicodeError):
+        return None
+    return parse_verdict(text).canonical
+
+
 def _run_gate_validation(run_root: Path) -> dict[str, Any]:
     """Run the canonical validator over one run directory."""
     spec = run_root / "spec.md"
@@ -586,7 +614,7 @@ def _decision_records_from_run(
 
 
 def _render_index_html(
-    payload: dict[str, Any], matrix: Any, snap_dir: Path
+    payload: dict[str, Any], matrix: Any, snap_dir: Path, locale: str
 ) -> str:
     """Render the delivery page with the payload inlined.
 
@@ -604,13 +632,22 @@ def _render_index_html(
             .replace('"', """)
         )
 
+    strings = STRINGS[locale]
+
+    def _display(value: Any) -> str:
+        if value is None:
+            return "—"
+        if isinstance(value, bool):
+            return strings["yes" if value else "no"]
+        return strings.get(str(value), str(value))
+
     statuses = payload.get("gateStatuses")
     rows = ""
     if isinstance(statuses, list):
         for index, state in enumerate(statuses[:8], start=1):
             rows += (
                 f'<tr><td>G{index}</td><td class="st st-{_esc(state)}">'
-                f"{_esc(state)}</td></tr>"
+                f"{_esc(_display(state))}</td></tr>"
             )
     viewports = payload.get("viewports")
     vp_rows = ""
@@ -625,9 +662,9 @@ def _render_index_html(
                 f"<td>{_esc(vp.get('name'))}</td>"
                 f"<td>{_esc(metrics.get('sw'))}</td>"
                 f"<td>{_esc(metrics.get('innerH'))}</td>"
-                f"<td>{_esc(metrics.get('hOverflow'))}</td>"
-                f"<td>{_esc(disclosure.get('inFold'))}</td>"
-                f"<td>{_esc(metrics.get('measurementStatus'))}</td>"
+                f"<td>{_esc(_display(metrics.get('hOverflow')))}</td>"
+                f"<td>{_esc(_display(disclosure.get('inFold')))}</td>"
+                f"<td>{_esc(_display(metrics.get('measurementStatus')))}</td>"
                 "</tr>"
             )
     decisions = payload.get("decisions")
@@ -638,7 +675,7 @@ def _render_index_html(
                 decision_items += (
                     f"<li><code>{_esc(item.get('id'))}</code> "
                     f"{_esc(item.get('title'))} "
-                    f"<small>({_esc(item.get('authority'))})</small></li>"
+                    f"<small>({_esc(_display(item.get('authority')))})</small></li>"
                 )
     payload_json = json.dumps(payload, ensure_ascii=False, indent=2)
     # A literal "</script>" inside the payload would close the block early;
@@ -651,27 +688,44 @@ def _render_index_html(
             matrix.get(vp), vp, snap_dir
         ):
             snap_figures += (
-                f'<figure><img src="snapshots/viewport-{vp}.png" alt="{vp} snapshot" '
+                f'<figure><img src="snapshots/viewport-{vp}.png" '
+                f'alt="{_esc(strings["snapshot"].format(viewport=vp))}" '
                 f'loading="lazy"/><figcaption>viewport-{vp}.png</figcaption></figure>'
             )
 
-    return (
-        _page_template()
-        .replace("%VERDICT%", _esc(payload.get("verdict")))
-        .replace("%VERDICT_LOWERCASE%", _esc(str(payload.get("verdict")).lower()))
-        .replace("%RUN_ID%", _esc(payload.get("runId")))
-        .replace("%AUTHORITY%", _esc(payload.get("authority")))
-        .replace("%PROFILE%", _esc(payload.get("profile")))
-        .replace("%TIMESTAMP%", _esc(payload.get("timestamp")))
-        .replace("%GATES_PASSED%", _esc(payload.get("gatesPassed")))
-        .replace("%CAPTURE_STATUS%", _esc(payload.get("captureStatus")))
-        .replace("%CONFIRMATION_NOTE%", _esc(payload.get("confirmationNote") or "—"))
-        .replace("%GATE_ROWS%", rows)
-        .replace("%VIEWPORT_ROWS%", vp_rows)
-        .replace("%DECISION_ITEMS%", decision_items)
-        .replace("%SNAPSHOT_FIGURES%", snap_figures)
-        .replace("%PAYLOAD_JSON%", payload_json)
-    )
+    note_key = "confirmed_note" if payload.get("authority") == "confirmed-user" else "pending_note"
+    note = _esc(strings[note_key])
+    if payload.get("confirmationNote"):
+        # Runtime diagnostics remain verbatim, distinct from localized UI copy.
+        note += (
+            f'<details><summary>{_esc(strings["diagnostic"])}</summary>'
+            f'<pre>{_esc(payload["confirmationNote"])}</pre></details>'
+        )
+    values = {key.upper(): _esc(text) for key, text in strings.items()}
+    values.update({
+        "LANG": locale,
+        "VERDICT_LABEL": strings["verdict"],
+        "AUTHORITY_LABEL": strings["authority"],
+        "PROFILE_LABEL": strings["profile"],
+        "GATES_PASSED_LABEL": strings["gates_passed"],
+        "VERDICT": _esc(_display(payload.get("verdict"))),
+        "VERDICT_LOWERCASE": _esc(str(payload.get("verdict")).lower()),
+        "RUN_ID": _esc(payload.get("runId")),
+        "AUTHORITY": _esc(_display(payload.get("authority"))),
+        "PROFILE": _esc(_display(payload.get("profile"))),
+        "TIMESTAMP": _esc(payload.get("timestamp")),
+        "GATES_PASSED": _esc(payload.get("gatesPassed")),
+        "CAPTURE_STATUS": _esc(_display(payload.get("captureStatus"))),
+        "CONFIRMATION_NOTE": note,
+        "GATE_ROWS": rows,
+        "VIEWPORT_ROWS": vp_rows,
+        "DECISION_ITEMS": decision_items,
+        "SNAPSHOT_FIGURES": snap_figures,
+        "PAYLOAD_JSON": payload_json,
+        "PAYLOAD_TEXT": _esc(json.dumps(payload, ensure_ascii=False, indent=2)),
+    })
+    # One pass: user-authored text resembling a template marker stays literal.
+    return re.sub(r"%([A-Z_]+)%", lambda match: values[match[1]], _page_template())
 
 
 def build_static_handoff(
@@ -683,6 +737,7 @@ def build_static_handoff(
     capture_runner: Callable[..., Any] | None = None,
     gate_runner: Callable[..., Any] | None = None,
     out_dir: Path | None = None,
+    locale: str | None = None,
 ) -> StaticHandoffResult:
     """Build the Stage 9 static handoff from durable run artifacts.
 
@@ -691,8 +746,10 @@ def build_static_handoff(
     output (``filled-ui.html``) - the page the five-viewport matrix and the
     layout probe actually target (ADR-0034 §4). Everything is written under
     ``<run_root>/evidence/static-handoff/``: snapshots, the disclosure JSON,
-    the ZIP package, and a self-contained index page.
+    the ZIP package, a same-directory ``deliverable.html`` copy (the page's
+    relative link target, spec A5), and a self-contained index page.
     """
+    locale = resolve_ui_locale(locale)
     run_root = Path(run_root)
     deliverable = Path(deliverable)
     out_dir = Path(out_dir) if out_dir is not None else run_root / "evidence" / "static-handoff"
@@ -707,6 +764,14 @@ def build_static_handoff(
     if gate_runner is None:
         gate_runner = _run_gate_validation
 
+    # Read the deliverable source up front and fail fast: its bytes are the
+    # run identity (hash), the ZIP's prototype member, and the on-disk copy
+    # the delivery page links relatively (#107). A missing or undecodable
+    # source aborts before any capture launches or artifact is written, so a
+    # delivery page can never exist without its link target.
+    deliverable_bytes = deliverable.read_bytes()
+    deliverable_text = deliverable_bytes.decode("utf-8")
+
     # Sample conditional-gate preconditions BEFORE writing anything: this
     # builder's own output lives under evidence/, and a precondition sampled
     # afterwards would be one this run manufactured for itself.
@@ -762,14 +827,25 @@ def build_static_handoff(
     confirmed, confirmation_reason = _confirmation_from_record(run_root, round_n)
 
     capture_complete = capture_status == "captured"
-    if confirmed and capture_complete and gate.get("available") and gates_resolved:
+    # The point-back owner verdict gates a delivery Pass (decision 25): a
+    # confirmed round with resolved gates still cannot convert a
+    # Recirculate/blocked owner verdict - or a missing/ambiguous one - into
+    # Pass. Pending/Recirculate behaviour and gatesPassed semantics are
+    # unchanged.
+    owner_pass = _point_back_verdict(run_root) == "pass"
+    if (
+        confirmed
+        and owner_pass
+        and capture_complete
+        and gate.get("available")
+        and gates_resolved
+    ):
         verdict = "Pass"
     elif confirmed:
         verdict = "Recirculate"
     else:
         verdict = "Pending"
 
-    deliverable_bytes = deliverable.read_bytes()
     run_id = (
         f"static-handoff-{round_n}-{hashlib.sha256(deliverable_bytes).hexdigest()[:12]}"
     )
@@ -808,6 +884,13 @@ def build_static_handoff(
     json_path = out_dir / "disclosure-review.json"
     json_path.write_text(disclosure_json(payload), encoding="utf-8")
 
+    # The delivery page links "deliverable.html" as a same-directory relative
+    # anchor (spec A5: disk artifacts, same-directory relative links); the copy
+    # must exist beside index.html, byte-identical to the ZIP member, or the
+    # delivery surface ships a dead link (#107).
+    deliverable_copy = out_dir / "deliverable.html"
+    deliverable_copy.write_bytes(deliverable_bytes)
+
     artifacts: dict[str, str] = {}
     if snap_dir.is_dir():
         for vp in VIEWPORT_ORDER:
@@ -820,13 +903,13 @@ def build_static_handoff(
         artifact_files=artifacts,
         # spec §4.1: the handoff ships "snapshots and prototype code". PNGs
         # alone do not let the recipient rebuild the reviewed page.
-        text_members={"deliverable.html": deliverable_bytes.decode("utf-8")},
+        text_members={"deliverable.html": deliverable_text},
         zip_target=str(zip_path),
     )
 
     index_html = out_dir / "index.html"
     index_html.write_text(
-        _render_index_html(payload, matrix, snap_dir), encoding="utf-8"
+        _render_index_html(payload, matrix, snap_dir, locale), encoding="utf-8"
     )
 
     return StaticHandoffResult(
@@ -835,4 +918,5 @@ def build_static_handoff(
         json_path=json_path,
         zip_path=zip_path,
         index_html=index_html,
+        deliverable_html=deliverable_copy,
     )