feat(explore): enforce scoped composition lineage and lifecycle evidence

apps/presentation/dashboard/src/features/personal-workspace/capability-localization.ts

@@ -37,7 +37,7 @@ const capabilityCopy: Record<WorkspaceLocale, Record<string, LocalizedCopy>> = {
     },
     explore_harness: {
       displayName: "Explore Harness",
-      description: "Keeps an evidence graph of exploration, with optional branch planning. Planning includes evidence; worker permissions remain separate.",
+      description: "Keeps an exploration evidence graph with optional branch planning. Explicit composition replans require a bound experiment or typed result; worker permissions and task completion remain separate.",
     },
     lark_event_inbox: {
       displayName: "Lark event inbox",
@@ -104,7 +104,7 @@ const capabilityCopy: Record<WorkspaceLocale, Record<string, LocalizedCopy>> = {
     },
     explore_harness: {
       displayName: "探索 Harness",
-      description: "记录探索证据图谱，可选开启分支规划；规划自动配套证据层，派生 Agent 权限独立。",
+      description: "记录探索证据图谱，可选开启分支规划。显式组合的重新规划需绑定实验或类型化结果；派生 Agent 权限与任务完成独立判断。",
     },
     lark_event_inbox: {
       displayName: "飞书事件收件箱",
@@ -167,6 +167,8 @@ const fieldCopy: Record<WorkspaceLocale, FieldCopy> = {
     reasoning_effort: { label: "Child reasoning effort", description: "For example max; the host must support this model and effort." },
     max_children: { label: "Maximum children", description: "Hard upper bound for concurrently delegated child work." },
     profile: { label: "Planner profile", description: "Select one registered Explore Harness profile." },
+    composition_mode: { label: "Composition policy", description: "Replan requires an exact experiment successor or typed result; it grants no execution authority.", options: {disabled: "Off", explicit_only: "Explicit candidates"} },
+    composition_scope_id: { label: "Research coverage scope", description: "An opaque scope id required by the explicit-only composition policy." },
     profile_preset: { label: "Report profile", description: "Capability-owned report profile, such as weekly-progress." },
     wait_for_ci: { label: "Wait for CI", description: "Disable to use local validation without querying or waiting for CI. Merge authority is unchanged." },
     review_order: { label: "Review direction", description: "Forward visits other authors first and oldest first. Reverse inverts the whole actionable queue." },
@@ -199,6 +201,8 @@ const fieldCopy: Record<WorkspaceLocale, FieldCopy> = {
     reasoning_effort: { label: "子 Agent 推理档位", description: "例如 max；宿主须支持所选模型与档位。" },
     max_children: { label: "最大子 Agent 数", description: "可同时委派的子任务硬上限。" },
     profile: { label: "规划 Profile", description: "选择一个已注册的 Explore Harness profile。" },
+    composition_mode: { label: "组合策略", description: "重新规划需要精确的实验后继或类型化结果；它不授予执行权限。", options: {disabled: "关闭", explicit_only: "仅显式候选"} },
+    composition_scope_id: { label: "研究覆盖范围", description: "显式组合策略要求填写不含私有内容的范围标识。" },
     profile_preset: { label: "报告 Profile", description: "由该能力管理的报告 profile，例如 weekly-progress。" },
     wait_for_ci: { label: "等待 CI", description: "关闭后使用本地验证，不查询或等待 CI；不改变合并权限。" },
     review_order: { label: "审阅方向", description: "正向先审其他作者，同层先审较早就绪的 PR；反向将整个可执行队列倒序。" },

apps/presentation/dashboard/src/features/personal-workspace/goal-capability-settings.tsx

@@ -1,4 +1,4 @@
-import { useEffect, useMemo, useState } from "react";
+import { useEffect, useLayoutEffect, useMemo, useState } from "react";
 import { AlertTriangle, Code2, LoaderCircle, RefreshCw } from "lucide-react";
 
 import {
@@ -72,7 +72,9 @@ function useCapabilityMutation({ goalId, onApplied, selected, t }: Readonly<{
     ? parseEditableCapabilityJson(selected.configuration_editor, jsonDraft) : null, [selected, jsonDraft]);
   const jsonValid = editorMode === "guided" || parsedJson !== null;
 
-  useEffect(() => {
+  // Initialize the new capability's draft before it can receive input. A
+  // post-paint reset can otherwise erase the first toggle after selection.
+  useLayoutEffect(() => {
     setEditorMode("guided");
     setJsonDraft("");
     const current = (selected?.capability_id === "pull_request_review" ? selected.effective_configuration?.configuration : selected?.current)

docs/architecture/rfcs/research-exploration-control-plane-v0.md

@@ -151,11 +151,11 @@ milestone status.
 
 | Gap | Consequence |
 |---|---|
-| No shared research write-time gate | The cold evidence codec cannot discharge or enforce a live composition obligation. |
-| No exact obligation/Todo/result lineage | A research receipt is not proof of an authorized Todo transition or accepted Goal closure. |
-| Cold shadow not adopted by hot status/frontier | The existing #3173 projection remains behavior-compatible; canonical research obligations still need M3 integration. |
-| No dismissal or deferral contract | Evidence invalidation is visible, but typed candidate retirement and resumption remain unimplemented. |
-| No experiment-result adoption gate | Negative evidence can be recorded without proving that the next bound decision used its scope, validity or uncertainty. Proposed single-experiment integration is defined in §11.5; current cold receipts do not enforce it. |
+| Mainline M3 write-time gate and exact lineage are not integrated | Open PR #5280 proposes an exact obligation/Todo/experiment/input gate. Its local receipt tests do not establish accepted Goal closure, and current-head CI, mergeability and maintainer effect acceptance remain open. |
+| Cold shadow not adopted by hot status/frontier on main | The existing #3173 projection remains behavior-compatible; the proposed M3 consumer adoption still needs integrated entrypoint qualification. |
+| Dismissal and deferral are not integrated | PR #5280 locally implements typed candidate dismissal, blocker waits and exact invalidated-duty retirement; review and mainline readback remain separate gates. |
+| No experiment-result adoption gate | Negative evidence can be recorded without proving that the next bound decision used its scope, validity or uncertainty. Proposed single-experiment integration is defined in §11.5; current cold receipts and PR #5280 do not enforce it. |
+| Execution attribution is not effect authority | An exact Todo/experiment/input receipt does not grant a task lease, effect permission or accepted Goal closure. |
 | Live qualification incomplete | Deterministic and real CLI/file-log tests establish state semantics, not model selection quality or scientific truth; no live Lark sync is qualified by projection tests. |
 | No promotion evidence for inferred combinations | Shared constraints are not known to be precise enough to trigger obligations. |
 
@@ -234,8 +234,9 @@ execution, and result into one ambiguous relation. This RFC rejects that shape.
 
 The research envelope and closure basis have an active CLI caller and the
 [versioned evidence protocol](../../reference/protocols/research-observation-v0.md).
-The action signature, shared write gate and model selection below remain design
-targets. The cold shadow does not promote them into current behavior.
+The opt-in M3 development path implements exact execution lineage and shared
+write gates and bounded retirement transitions. Model selection below
+remain design targets; the cold shadow does not activate enforcement.
 
 ### 7.1 Compose; do not mutate v0 silently
 
@@ -919,7 +920,7 @@ control-plane failures.
 | M0 | RFC, current-state inventory, and explicit ownership decision | Maintainer review; no runtime behavior | Accepted design |
 | M1 | Characterization fixtures plus typed research observation and closure contract in Explore | Deterministic normalization, privacy, compatibility, and negative tests | Implemented evidence/CLI slice; live research qualification remains separate |
 | M2 | Explicit-only composition candidate, canonical gap projection, and read-only status shadow | No pairwise inference; bounded packet; projection parity | Partial: #3173 legacy quota/successor; canonical binary cold shadow in CLI/Lark projection; hot status adoption and live Lark qualification remain |
-| M3 | Goal-frontier obligation, exact Todo/experiment lineage, shared write-time gate; proposed §11.5 single-experiment result/adoption prerequisite | State/replay matrix, real provider/state readback, default-off parity and packaged journey before enforcement | Not started; §11.5 is a design proposal, not delivered runtime |
+| M3 | Goal-frontier obligation, exact Todo/experiment lineage, shared write-time gate; proposed §11.5 single-experiment result/adoption prerequisite | State/replay matrix, real provider/state readback, default-off parity and packaged journey before enforcement | Proposed in open PR #5280, not integrated: local state/replay and premerge checks exist, while current-head CI, mainline conflict, public effect case and maintainer acceptance remain open. §11.5 is a design proposal, not delivered runtime |
 | M4 | Bounded multi-candidate cards, `composition_selection_v0`, real model-tool behavior qualification, and repeated live shadow | Model autonomously selects a legal semantic action from the delivered candidate set; selection quality is no worse than the declared fallback; compact receipts only | Not started |
 | M5 | Shared-constraint candidate ranking in shadow mode | Precision and cost evidence; no automatic trigger | Not started |
 | M6 | Optional inferred trigger | Explicit maintainer decision and measured promotion thresholds | Deferred |
@@ -954,6 +955,33 @@ M3 is the first behavior-changing slice. It should be a separate PR so the
 obligation and write gate can be reviewed and reverted independently from the
 evidence schema.
 
+The M3 development boundary joins current same-agent Todo, experiment and input
+facts in the typed Explore owner. Goal policy is explicitly scoped and disabled
+by default; the existing capability editor and CLI share its configuration
+owner. Quota and refresh reuse one live frontier, with the original duty pinned
+through scalar rollout fields and both receipt adapters. Real CLI tests reject
+unrelated/deferred successors and invalidated writeback, then settle the original
+Turn through its exact successor. File/SQLite tests distinguish canonical Todos
+from stale display rows; packaged UI checks exercise policy preview, activation,
+disable, readback and narrow screens. Native actor/lease and CAS admission remain in force; a fixed IO host
+locks the graph while the typed owner qualifies and persists completion evidence.
+Real File/SQLite and legacy tests cover missing evidence, direct IPC self-approval,
+terminal-verb bypass, retained archive lineage and immutable completion replay.
+Agent-scoped status, Explore and existing Lark Summary fields use the same live
+facts. Typed candidate dismissal permits scoped terminal retirement; a fresh
+canonical blocker and common Todo resume condition defer the gap without closing
+it. Real CLI validates exact observed/dismissed/blocked progress source through
+the original Turn, with independent File/SQLite lease-safe wait/resume evidence.
+Invalidated input/scope/activation produces source-qualified retirement for the
+original duty. Common readback retains its guard and historical debit, closes
+that Turn without spend, and keeps the current frontier and runnable/paused work
+visible. IO assembly stays in existing CLI/refresh composition roots; the shared
+gate consumes supplied capability facts and the typed owner remains singular.
+The local state/replay matrix and 19 standard risk-selected premerge checks pass.
+Two existing scheduler ACK tests fail identically on the unchanged base under the
+same local runtime; this is retained as a baseline limitation, not called green.
+Maintainer-reviewed integration and independent live qualification remain.
+
 ## 17. Rejected Alternatives
 
 ### 17.1 Automatically pair nodes with a shared closure stage

docs/architecture/rfcs/research-exploration-control-plane-v0.zh-CN.md

@@ -132,11 +132,11 @@ hypothesis”或“新 probe family”。它无法持久表达：A 和 B 都已
 
 | 缺口 | 后果 |
 |---|---|
-| 没有 shared research write-time gate | Cold evidence codec 不能解除或强制 live composition obligation。 |
-| 没有精确 obligation/Todo/result lineage | Research receipt 不证明已授权 Todo transition 或已接受 Goal closure。 |
-| Cold shadow 未接入 hot status/frontier | 既有 #3173 投影保持行为兼容；canonical research obligation 仍需 M3 集成。 |
-| 没有 dismissal/deferral contract | Evidence invalidation 可见，但类型化 candidate retirement/resumption 尚未实现。 |
-| 没有 experiment-result adoption gate | 负证据可被记录，但不能证明下一绑定决策使用了其 scope、validity 与 uncertainty。§11.5 提议 single-experiment 集成；当前 cold receipt 不强制执行该提案。 |
+| Mainline 尚未集成 M3 写入门禁与精确 lineage | 开放的 PR #5280 提议精确 obligation/Todo/experiment/input 门禁；本地 receipt 测试不证明 Goal 已接受关闭，当前 head 的 CI、可合并性与维护者效果验收仍未完成。 |
+| Cold shadow 尚未在 main 接入 hot status/frontier | 既有 #3173 投影保持行为兼容；提议中的 M3 消费端仍需集成后的入口资格验证。 |
+| Dismissal 与 deferral 尚未集成 | PR #5280 在本地实现 typed candidate dismissal、blocker wait 与精确 invalidated-duty retirement；review 和主干回读仍是独立门槛。 |
+| 尚无 experiment-result adoption gate | 负证据可被记录，但不能证明下一绑定决策使用了其 scope、validity 与 uncertainty。§11.5 提议 single-experiment 集成；当前 cold receipt 与 PR #5280 均未强制执行。 |
+| 执行 attribution 不是 effect 权限 | 精确 Todo/experiment/input receipt 不授予 task lease、effect 权限或已接受的 Goal closure。 |
 | Live qualification 不完整 | Deterministic 与真实 CLI/file-log 测试证明状态语义，不证明 model selection 质量或科学结论；projection 测试不构成 live Lark sync 资格。 |
 | inferred combination 没有 promotion evidence | 共享 constraint 的精度还不足以直接触发 obligation。 |
 
@@ -211,8 +211,8 @@ A、B 之间的 `joint_probe` 直连边会把 candidate、execution 和 result 
 
 Research envelope 与 closure basis 已有真实 CLI caller 和
 [版本化证据协议](../../reference/protocols/research-observation-v0.zh-CN.md)。
-下文 action signature、shared write gate 和 model selection 仍为设计目标；
-cold shadow 不会将其 promotion 为当前行为。
+Opt-in M3 开发路径已实现精确 execution lineage 和共享 write gate。下文 model
+selection 仍为设计目标；有界 retirement 已实现，cold shadow 不会激活门禁。
 
 ### 7.1 组合，而不是静默修改 v0
 
@@ -839,7 +839,7 @@ rule，以及 model variance 与 control-plane failure 的分离。
 | M0 | RFC、current-state inventory 与显式 ownership decision | Maintainer review；无 runtime behavior | 已接受的设计 |
 | M1 | Characterization fixture，以及 Explore 中的 typed research observation 与 closure contract | Deterministic normalization、privacy、compatibility 与 negative test | Evidence/CLI 切片已实现；真实研究 qualification 独立保留 |
 | M2 | Explicit-only composition candidate、canonical gap projection 与 read-only status shadow | 不做 pairwise inference；packet 有界；projection parity | 部分实现：#3173 legacy quota/successor；CLI/Lark projection 的 canonical binary cold shadow；hot status adoption 与 live Lark qualification 仍未完成 |
-| M3 | Goal-frontier obligation、精确 Todo/experiment lineage、共享 write-time gate；§11.5 提议 single-experiment result/adoption prerequisite | State/replay matrix、真实 provider/state readback、default-off parity；enforcement 前验证 packaged journey | 未开始；§11.5 是设计提案，不是已交付 runtime |
+| M3 | Goal-frontier obligation、精确 Todo/experiment lineage、共享 write-time gate；§11.5 提议 single-experiment result/adoption prerequisite | State/replay matrix、真实 provider/state readback、default-off parity；enforcement 前验证 packaged journey | 开放 PR #5280 提案，尚未集成：已有本地 state/replay 与 premerge 检查，但当前 head 的 CI、主干冲突、公开效果案例与维护者验收仍未闭合。§11.5 是设计提案，不是已交付 runtime |
 | M4 | 有界 multi-candidate card、`composition_selection_v0`、真实 model-tool behavior qualification 与重复 live shadow | 模型从交付 candidate set 中自主选择合法 semantic action；选择质量不劣于 declared fallback；只保留 compact receipt | 未开始 |
 | M5 | Shared-constraint candidate 在 shadow mode 中排序 | 有 precision/cost evidence；不自动触发 | 未开始 |
 | M6 | 可选 inferred trigger | 显式 maintainer decision 与量化 promotion threshold | 延后 |
@@ -872,6 +872,27 @@ milestone gap，不另建平行 task tree。
 M3 是第一个 behavior-changing slice。它应单独成 PR，使 obligation 与 write gate
 能够独立于 evidence schema 评审和回滚。
 
+M3 开发边界在 typed Explore owner 中连接当前同一 Agent 的 Todo、experiment
+和 input 事实。Goal policy 显式限定范围且默认关闭；既有能力编辑器与 CLI 共享
+配置 owner。Quota 与 refresh 复用同一 live frontier，原 duty 通过 rollout 标量
+字段和两端 receipt adapter 固定。真实 CLI 测试拒绝无关/延期 successor 与失效
+writeback，再通过精确 successor 结算原 Turn。File/SQLite 测试区分 canonical
+Todo 与陈旧显示行；打包 UI 验证策略预览、启用、关闭、读回和窄屏。Native actor/lease 与 CAS admission
+仍强制执行；固定 IO host 锁定图，typed owner 校验并持久化 completion evidence。
+真实 File/SQLite 与 legacy 测试覆盖缺少证据、direct IPC 自报 approval、terminal
+verb 绕过、保留 archive lineage 和不可变 completion 回放。Agent 范围的 status、
+Explore 与既有 Lark Summary 字段使用同一 live fact。Typed candidate dismissal
+允许 scoped terminal retirement；新 canonical blocker 与通用 Todo resume condition
+使 gap 暂缓，但不关闭。真实 CLI 通过原 Turn 校验 observed/dismissed/blocked 的
+精确 progress source；File/SQLite 独立验证 lease-safe wait/resume。
+输入/scope/activation 失效时，为原 duty 生成 source-qualified retirement。
+通用 readback 保留原 guard 与历史 debit，无支出关闭该 Turn，同时保持当前
+frontier 与 runnable/paused work 可见。IO 装配保留在既有 CLI/refresh composition
+root；共享门禁消费传入的 capability fact，typed owner 始终只有一个。
+本地 state/replay matrix 与 19 项 standard 风险验证通过。两个既有 scheduler ACK
+测试在相同本地 runtime、未修改 base 上也以相同方式失败；保留为 baseline 限制，
+不称为 green。Maintainer review 集成与独立 live qualification 仍未完成。
+
 ## 17. 被拒绝的替代方案
 
 ### 17.1 自动组合拥有共享 closure stage 的 node