@boyingliu01/xp-gate 0.10.6 → 0.10.7

package/mock-policy/AGENTS.md

@@ -1,9 +1,9 @@
 # SRC/MOCK-POLICY KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Mock layering policy enforcement — Gate M3 of pre-push hook. Ensures integration tests use real implementations for internal dependencies, mock external dependencies, and annotate pending mocks with removal plans. Combines project scope scanning, mock decision engine, and per-file validation into a single pipeline.

package/mutation/AGENTS.md

@@ -1,9 +1,9 @@
 # SRC/MUTATION KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 **Gate M** (incremental mutation testing) + **Gate M2** helpers (test-layer detection used by `src/mock-policy/`). Pre-push quality gate. TypeScript-only; uses Stryker.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@boyingliu01/xp-gate",
-  "version": "0.10.6",
+  "version": "0.10.7",
   "description": "AI-driven development workflow: 6 quality gates + Delphi review + Sprint Flow",
   "bin": {
     "xp-gate": "./bin/xp-gate.js"
@@ -18,6 +18,9 @@
     "mock-policy/",
     "build-integrity/"
   ],
+  "exports": {
+    "./tui": "./plugins/opencode/tui-plugin.ts"
+  },
   "repository": {
     "type": "git",
     "url": "https://github.com/boyingliu01/xp-gate"

package/plugins/claude-code/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "xp-gate",
-  "version": "0.10.6",
+  "version": "0.10.7",
   "displayName": "XP-Gate",
   "description": "Extreme Programming quality gates + AI workflow skills for Claude Code. Includes 10 quality gates (Gate 0-9), Sprint Flow (11 phases), and Delphi multi-expert review (>=90% consensus).",
   "author": {

package/plugins/claude-code/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, 3 experts from ≥2 providers, domestic models only). Supports design + code-walkthrough modes.

package/plugins/claude-code/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 **11-phase** development pipeline: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW → USER ACCEPTANCE → FEEDBACK → SHIP → LAND → CLEANUP. Phase 2 default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE in Phase 1: design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/claude-code/skills/sprint-flow/SKILL.md

@@ -231,13 +231,17 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 4. **所有路由必须产生** `delphi-reviewed.json` (verdict: APPROVED) 才能进入 Phase 2 BUILD
 
 ### Phase 0: THINK（需求探索与设计）
-- **Subagent dispatch**: orchestrator 通过 `task(category="deep", load_skills=["brainstorming"])` 启动独立 session
+- **Orchestrator 直接执行**: brainstorming 是交互式 skill，**必须由 orchestrator 直接调用** `skill(name="brainstorming")`，不可 dispatch 到 subagent（Issue #217, #225, #248）
 - 输入: Phase -1 summary（worktree 路径）+ 用户原始需求
 - 输出: 结构化设计文档 → 直接作为 Phase 1 PLAN 的输入
 - **HARD-GATE**: 设计未批准 → 不可进入实现
 
 ### Phase 1: PLAN（共识评审）
-- **Subagent dispatch**: orchestrator 通过 `task(category="deep", load_skills=["autoplan", "delphi-review", "to-issues"])` 启动独立 session
+- **注意**: `autoplan` 和 `to-issues` 是交互式 skill，**必须由 orchestrator 直接执行**（Issue #225, #248）
+- **执行模式（三阶段）**:
+  1. **Orchestrator 直接执行 autoplan**: `skill(name="autoplan")` → 用户确认 taste_decisions
+  2. **Orchestrator 直接执行 to-issues**: `skill(name="to-issues")` → 用户确认 Issue 拆分方案
+  3. **Subagent 执行 delphi-review**: `task(category="deep", load_skills=["delphi-review"])` — 非交互式，可自动运行至 APPROVED
 - 输入: phase-0-summary.md + 设计文档
 - 输出: `specification.yaml`（含 user_stories[]）+ `slices-manifest.json`
 
@@ -304,7 +308,7 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 **详细指令**: 参见 `references/phase-6-ship.md` — GITHOOKS-GATE / VERSION-GATE / VERSION CHANGESET (Issue #142) / changeset schema。
 
 **快速参考**:
-- **Dispatch**: `task(category="quick", load_skills=["finishing-a-development-branch", "ship"])`
+- **Orchestrator 直接执行**: `finishing-a-development-branch` 和 `ship` 均为交互式 skill（4 选项菜单 + PR 确认），**必须由 orchestrator 直接调用** `skill(name="finishing-a-development-branch")` 和 `skill(name="ship")`
 - 输入: phase-5-summary.md + feedback-log.md → 输出: PR URL
 - **HARD-GATE**: Phase 5 未完成 → BLOCK。验证 `feedback-log.md` 存在。
 - **GITHOOKS-GATE**: 验证 hooks 完整性，缺失则 `githooks/install.sh`
@@ -315,7 +319,7 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 **详细指令**: 参见 `references/phase-7-land.md` — 完整流程、SLA 指标、回滚策略。
 
 **快速参考**:
-- **Dispatch**: `task(category="deep", load_skills=["land-and-deploy"])`
+- **Orchestrator 直接执行**: `land-and-deploy` 包含 merge 确认和 rollback 决策，**必须由 orchestrator 直接调用** `skill(name="land-and-deploy")`
 - 输入: phase-6-summary.md + PR URL → 输出: 部署状态 + Canary 报告
 - 流程: Merge PR → 等待 CI (10min) → 等待 Deploy (10min) → Canary Health Check (5min)
 - **回滚**: `git revert` 最后一次 merge commit

package/plugins/claude-code/skills/sprint-flow/references/orchestration-rules.md

@@ -30,16 +30,32 @@
 |-------|------|:---------:|----------|-------------|--------|
 | -1 | ISOLATE | ❌ | Bash（直接执行） | 无 | orchestrator |
 | -0.5 | AUTO-ESTIMATE | ❌ | Bash（直接执行） | 无 | orchestrator |
-| 0 | THINK | ✅ | `deep` | `["brainstorming"]` | subagent |
-| 1 | PLAN | ✅ | `deep` | `["autoplan", "delphi-review", "to-issues"]` | subagent |
-| 2 | BUILD | ✅(已有) | ralph-loop | `["test-driven-development"]` | subagent |
+| 0 | THINK | ❌ | orchestrator（直接执行） | `["brainstorming"]` | orchestrator |
+| 1 | PLAN | ⚠️ | orchestrator 执行 autoplan + to-issues（交互）→ subagent 执行 delphi-review | 见下方说明 | orchestrator + subagent |
+| 2 | BUILD | ✅ | ralph-loop | `["test-driven-development"]` | subagent |
 | 3 | REVIEW | ✅ | `deep` | `["delphi-review", "test-specification-alignment"]` | subagent |
 | 4 | USER ACCEPT | ❌ | **强制人工** | 无 | 用户 |
 | 5 | FEEDBACK | ✅ | `quick` | `["learn", "retro", "systematic-debugging"]` | subagent |
-| 6 | SHIP | ✅ | `quick` | `["finishing-a-development-branch", "ship"]` | subagent |
-| 7 | LAND | ✅ | `deep` | `["land-and-deploy"]` | subagent |
+| 6 | SHIP | ❌ | orchestrator（直接执行） | `["finishing-a-development-branch", "ship"]` | orchestrator |
+| 7 | LAND | ❌ | orchestrator（直接执行） | `["land-and-deploy"]` | orchestrator |
 | 8 | CLEANUP | ❌ | Bash（直接执行） | 无 | orchestrator |
 
+**⚠️ 交互式 skill 必须由 orchestrator 直接执行（不可 dispatch 到 subagent）**：
+
+| Phase | Skill | 为什么不能在 subagent 中执行 |
+|-------|-------|---------------------------|
+| 0 | `brainstorming` | 需要与用户对话确认需求、提出澄清问题。Subagent 是 fire-and-forget 模式，无法暂停等待用户输入（Issue #217） |
+| 1 | `autoplan` | taste_decisions 节点暂停等待用户确认。必须由 orchestrator 直接执行（Issue #225） |
+| 1 | `to-issues` | Step 6 "向用户确认" — 展示拆分结果，等待用户批准后才生成 slices-manifest.json |
+| 6 | `finishing-a-development-branch` | 4 选项菜单 (merge/PR/keep/discard) 需要用户选择；Option 4 (discard) 要求 typed confirmation |
+| 6 | `ship` | PR 创建前需要用户确认；包含 AskUserQuestion STOP 点 |
+| 7 | `land-and-deploy` | Merge 确认、rollback 决策 — 均为用户交互点 |
+
+**Phase 1 执行模式（两阶段）**：
+1. **Orchestrator 直接执行 autoplan**：`skill(name="autoplan")` → 等待用户确认 taste_decisions
+2. **Orchestrator 直接执行 to-issues**：`skill(name="to-issues")` → 等待用户确认 Issue 拆分
+3. **Subagent 执行 delphi-review**：`task(category="deep", load_skills=["delphi-review"])` — 非交互式，可自动运行至 APPROVED
+
 **上下文隔离原则**：
 - 每个 Subagent 在**独立 session** 中启动，不继承 orchestrator 的对话历史
 - orchestrator session 仅接收 subagent 的最终结果摘要（~13,000 tokens/sprint）
@@ -98,6 +114,20 @@
    - 时机：每个 Phase 完成后的 transition 阶段自动展示，用户无需请求
    - 向后兼容：旧版 `sprint-state.json` 缺少 `phase_history` 时，从 `phase` 字段推断状态
 
+### Background Task Resume Protocol (MANDATORY — Issue #248)
+
+After dispatching background agents with `run_in_background=true` (e.g., Phase -0.5 AUTO-ESTIMATE explore agents, Phase 1 delphi-review subagent):
+
+1. **END YOUR RESPONSE** — do NOT poll `background_output()` before receiving `<system-reminder>`
+2. **On `<system-reminder>` notification**: collect ALL completed results via `background_output(task_id="bg_...")`
+3. **When ALL tasks complete**: immediately resume by:
+   a. Synthesizing results into the current phase assessment
+   b. Executing the phase transition gate
+   c. Continuing to the next phase **WITHOUT waiting for human input**
+4. **If any task failed**: collect partial results, log the failure, continue with available data (do NOT block the pipeline)
+
+**The orchestrator MUST treat task completion notifications as an implicit "continue" signal.** This protocol prevents the "human-in-loop stall" where all background tasks complete but the orchestrator waits indefinitely for a human message (Issue #235, #248).
+
 ### Phase Summary 格式（YAML Frontmatter Schema）
 
 每个 `phase-N-summary.md` 必须包含以下 YAML frontmatter：

package/plugins/claude-code/skills/sprint-flow/references/phase-minus-0-5-auto-estimate.md

@@ -228,6 +228,12 @@ git diff --stat HEAD 2>/dev/null  # 如果有局部修改
 
 ---
 
+## 编排注意事项
+
+Phase -0.5 经常使用并行 background explore agents 评估代码影响范围。Agent dispatch 后必须遵循 **Background Task Resume Protocol**（参见 `orchestration-rules.md`），在所有 background task 返回后自动恢复继续，不得等待人工消息（Issue #248）。
+
+---
+
 ## 与学习循环的集成
 
 ### 数据收集

package/plugins/claude-code/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/plugins/opencode/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@boyingliu01/opencode-plugin",
-  "version": "0.10.6",
+  "version": "0.10.7",
   "type": "module",
   "main": "index.ts",
   "description": "XP-Gate quality gates + AI workflow skills + Sprint Flow TUI sidebar for OpenCode",

package/plugins/opencode/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, 3 experts from ≥2 providers, domestic models only). Supports design + code-walkthrough modes.

package/plugins/opencode/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 **11-phase** development pipeline: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW → USER ACCEPTANCE → FEEDBACK → SHIP → LAND → CLEANUP. Phase 2 default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE in Phase 1: design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/opencode/skills/sprint-flow/SKILL.md

@@ -231,13 +231,17 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 4. **所有路由必须产生** `delphi-reviewed.json` (verdict: APPROVED) 才能进入 Phase 2 BUILD
 
 ### Phase 0: THINK（需求探索与设计）
-- **Subagent dispatch**: orchestrator 通过 `task(category="deep", load_skills=["brainstorming"])` 启动独立 session
+- **Orchestrator 直接执行**: brainstorming 是交互式 skill，**必须由 orchestrator 直接调用** `skill(name="brainstorming")`，不可 dispatch 到 subagent（Issue #217, #225, #248）
 - 输入: Phase -1 summary（worktree 路径）+ 用户原始需求
 - 输出: 结构化设计文档 → 直接作为 Phase 1 PLAN 的输入
 - **HARD-GATE**: 设计未批准 → 不可进入实现
 
 ### Phase 1: PLAN（共识评审）
-- **Subagent dispatch**: orchestrator 通过 `task(category="deep", load_skills=["autoplan", "delphi-review", "to-issues"])` 启动独立 session
+- **注意**: `autoplan` 和 `to-issues` 是交互式 skill，**必须由 orchestrator 直接执行**（Issue #225, #248）
+- **执行模式（三阶段）**:
+  1. **Orchestrator 直接执行 autoplan**: `skill(name="autoplan")` → 用户确认 taste_decisions
+  2. **Orchestrator 直接执行 to-issues**: `skill(name="to-issues")` → 用户确认 Issue 拆分方案
+  3. **Subagent 执行 delphi-review**: `task(category="deep", load_skills=["delphi-review"])` — 非交互式，可自动运行至 APPROVED
 - 输入: phase-0-summary.md + 设计文档
 - 输出: `specification.yaml`（含 user_stories[]）+ `slices-manifest.json`
 
@@ -304,7 +308,7 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 **详细指令**: 参见 `references/phase-6-ship.md` — GITHOOKS-GATE / VERSION-GATE / VERSION CHANGESET (Issue #142) / changeset schema。
 
 **快速参考**:
-- **Dispatch**: `task(category="quick", load_skills=["finishing-a-development-branch", "ship"])`
+- **Orchestrator 直接执行**: `finishing-a-development-branch` 和 `ship` 均为交互式 skill（4 选项菜单 + PR 确认），**必须由 orchestrator 直接调用** `skill(name="finishing-a-development-branch")` 和 `skill(name="ship")`
 - 输入: phase-5-summary.md + feedback-log.md → 输出: PR URL
 - **HARD-GATE**: Phase 5 未完成 → BLOCK。验证 `feedback-log.md` 存在。
 - **GITHOOKS-GATE**: 验证 hooks 完整性，缺失则 `githooks/install.sh`
@@ -315,7 +319,7 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 **详细指令**: 参见 `references/phase-7-land.md` — 完整流程、SLA 指标、回滚策略。
 
 **快速参考**:
-- **Dispatch**: `task(category="deep", load_skills=["land-and-deploy"])`
+- **Orchestrator 直接执行**: `land-and-deploy` 包含 merge 确认和 rollback 决策，**必须由 orchestrator 直接调用** `skill(name="land-and-deploy")`
 - 输入: phase-6-summary.md + PR URL → 输出: 部署状态 + Canary 报告
 - 流程: Merge PR → 等待 CI (10min) → 等待 Deploy (10min) → Canary Health Check (5min)
 - **回滚**: `git revert` 最后一次 merge commit

package/plugins/opencode/skills/sprint-flow/references/orchestration-rules.md

@@ -30,16 +30,32 @@
 |-------|------|:---------:|----------|-------------|--------|
 | -1 | ISOLATE | ❌ | Bash（直接执行） | 无 | orchestrator |
 | -0.5 | AUTO-ESTIMATE | ❌ | Bash（直接执行） | 无 | orchestrator |
-| 0 | THINK | ✅ | `deep` | `["brainstorming"]` | subagent |
-| 1 | PLAN | ✅ | `deep` | `["autoplan", "delphi-review", "to-issues"]` | subagent |
-| 2 | BUILD | ✅(已有) | ralph-loop | `["test-driven-development"]` | subagent |
+| 0 | THINK | ❌ | orchestrator（直接执行） | `["brainstorming"]` | orchestrator |
+| 1 | PLAN | ⚠️ | orchestrator 执行 autoplan + to-issues（交互）→ subagent 执行 delphi-review | 见下方说明 | orchestrator + subagent |
+| 2 | BUILD | ✅ | ralph-loop | `["test-driven-development"]` | subagent |
 | 3 | REVIEW | ✅ | `deep` | `["delphi-review", "test-specification-alignment"]` | subagent |
 | 4 | USER ACCEPT | ❌ | **强制人工** | 无 | 用户 |
 | 5 | FEEDBACK | ✅ | `quick` | `["learn", "retro", "systematic-debugging"]` | subagent |
-| 6 | SHIP | ✅ | `quick` | `["finishing-a-development-branch", "ship"]` | subagent |
-| 7 | LAND | ✅ | `deep` | `["land-and-deploy"]` | subagent |
+| 6 | SHIP | ❌ | orchestrator（直接执行） | `["finishing-a-development-branch", "ship"]` | orchestrator |
+| 7 | LAND | ❌ | orchestrator（直接执行） | `["land-and-deploy"]` | orchestrator |
 | 8 | CLEANUP | ❌ | Bash（直接执行） | 无 | orchestrator |
 
+**⚠️ 交互式 skill 必须由 orchestrator 直接执行（不可 dispatch 到 subagent）**：
+
+| Phase | Skill | 为什么不能在 subagent 中执行 |
+|-------|-------|---------------------------|
+| 0 | `brainstorming` | 需要与用户对话确认需求、提出澄清问题。Subagent 是 fire-and-forget 模式，无法暂停等待用户输入（Issue #217） |
+| 1 | `autoplan` | taste_decisions 节点暂停等待用户确认。必须由 orchestrator 直接执行（Issue #225） |
+| 1 | `to-issues` | Step 6 "向用户确认" — 展示拆分结果，等待用户批准后才生成 slices-manifest.json |
+| 6 | `finishing-a-development-branch` | 4 选项菜单 (merge/PR/keep/discard) 需要用户选择；Option 4 (discard) 要求 typed confirmation |
+| 6 | `ship` | PR 创建前需要用户确认；包含 AskUserQuestion STOP 点 |
+| 7 | `land-and-deploy` | Merge 确认、rollback 决策 — 均为用户交互点 |
+
+**Phase 1 执行模式（两阶段）**：
+1. **Orchestrator 直接执行 autoplan**：`skill(name="autoplan")` → 等待用户确认 taste_decisions
+2. **Orchestrator 直接执行 to-issues**：`skill(name="to-issues")` → 等待用户确认 Issue 拆分
+3. **Subagent 执行 delphi-review**：`task(category="deep", load_skills=["delphi-review"])` — 非交互式，可自动运行至 APPROVED
+
 **上下文隔离原则**：
 - 每个 Subagent 在**独立 session** 中启动，不继承 orchestrator 的对话历史
 - orchestrator session 仅接收 subagent 的最终结果摘要（~13,000 tokens/sprint）
@@ -98,6 +114,20 @@
    - 时机：每个 Phase 完成后的 transition 阶段自动展示，用户无需请求
    - 向后兼容：旧版 `sprint-state.json` 缺少 `phase_history` 时，从 `phase` 字段推断状态
 
+### Background Task Resume Protocol (MANDATORY — Issue #248)
+
+After dispatching background agents with `run_in_background=true` (e.g., Phase -0.5 AUTO-ESTIMATE explore agents, Phase 1 delphi-review subagent):
+
+1. **END YOUR RESPONSE** — do NOT poll `background_output()` before receiving `<system-reminder>`
+2. **On `<system-reminder>` notification**: collect ALL completed results via `background_output(task_id="bg_...")`
+3. **When ALL tasks complete**: immediately resume by:
+   a. Synthesizing results into the current phase assessment
+   b. Executing the phase transition gate
+   c. Continuing to the next phase **WITHOUT waiting for human input**
+4. **If any task failed**: collect partial results, log the failure, continue with available data (do NOT block the pipeline)
+
+**The orchestrator MUST treat task completion notifications as an implicit "continue" signal.** This protocol prevents the "human-in-loop stall" where all background tasks complete but the orchestrator waits indefinitely for a human message (Issue #235, #248).
+
 ### Phase Summary 格式（YAML Frontmatter Schema）
 
 每个 `phase-N-summary.md` 必须包含以下 YAML frontmatter：

package/plugins/opencode/skills/sprint-flow/references/phase-minus-0-5-auto-estimate.md

@@ -228,6 +228,12 @@ git diff --stat HEAD 2>/dev/null  # 如果有局部修改
 
 ---
 
+## 编排注意事项
+
+Phase -0.5 经常使用并行 background explore agents 评估代码影响范围。Agent dispatch 后必须遵循 **Background Task Resume Protocol**（参见 `orchestration-rules.md`），在所有 background task 返回后自动恢复继续，不得等待人工消息（Issue #248）。
+
+---
+
 ## 与学习循环的集成
 
 ### 数据收集

package/plugins/opencode/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/plugins/qoder/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, 3 experts from ≥2 providers, domestic models only). Supports design + code-walkthrough modes.

package/plugins/qoder/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 **11-phase** development pipeline: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW → USER ACCEPTANCE → FEEDBACK → SHIP → LAND → CLEANUP. Phase 2 default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE in Phase 1: design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/qoder/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/principles/AGENTS.md

@@ -1,9 +1,9 @@
 # PRINCIPLES CHECKER MODULE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Clean Code & SOLID principles checker — **Gate 4** of pre-commit. 14 rules × 9 language adapters, SARIF 2.1.0 output. Houses the **Boy Scout Rule** enforcement engine (Gate 6) and warning-baseline storage.

package/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, 3 experts from ≥2 providers, domestic models only). Supports design + code-walkthrough modes.

package/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 **11-phase** development pipeline: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW → USER ACCEPTANCE → FEEDBACK → SHIP → LAND → CLEANUP. Phase 2 default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE in Phase 1: design must pass Delphi review (≥90% consensus) before any coding.

package/skills/sprint-flow/SKILL.md

@@ -231,13 +231,17 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 4. **所有路由必须产生** `delphi-reviewed.json` (verdict: APPROVED) 才能进入 Phase 2 BUILD
 
 ### Phase 0: THINK（需求探索与设计）
-- **Subagent dispatch**: orchestrator 通过 `task(category="deep", load_skills=["brainstorming"])` 启动独立 session
+- **Orchestrator 直接执行**: brainstorming 是交互式 skill，**必须由 orchestrator 直接调用** `skill(name="brainstorming")`，不可 dispatch 到 subagent（Issue #217, #225, #248）
 - 输入: Phase -1 summary（worktree 路径）+ 用户原始需求
 - 输出: 结构化设计文档 → 直接作为 Phase 1 PLAN 的输入
 - **HARD-GATE**: 设计未批准 → 不可进入实现
 
 ### Phase 1: PLAN（共识评审）
-- **Subagent dispatch**: orchestrator 通过 `task(category="deep", load_skills=["autoplan", "delphi-review", "to-issues"])` 启动独立 session
+- **注意**: `autoplan` 和 `to-issues` 是交互式 skill，**必须由 orchestrator 直接执行**（Issue #225, #248）
+- **执行模式（三阶段）**:
+  1. **Orchestrator 直接执行 autoplan**: `skill(name="autoplan")` → 用户确认 taste_decisions
+  2. **Orchestrator 直接执行 to-issues**: `skill(name="to-issues")` → 用户确认 Issue 拆分方案
+  3. **Subagent 执行 delphi-review**: `task(category="deep", load_skills=["delphi-review"])` — 非交互式，可自动运行至 APPROVED
 - 输入: phase-0-summary.md + 设计文档
 - 输出: `specification.yaml`（含 user_stories[]）+ `slices-manifest.json`
 
@@ -304,7 +308,7 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 **详细指令**: 参见 `references/phase-6-ship.md` — GITHOOKS-GATE / VERSION-GATE / VERSION CHANGESET (Issue #142) / changeset schema。
 
 **快速参考**:
-- **Dispatch**: `task(category="quick", load_skills=["finishing-a-development-branch", "ship"])`
+- **Orchestrator 直接执行**: `finishing-a-development-branch` 和 `ship` 均为交互式 skill（4 选项菜单 + PR 确认），**必须由 orchestrator 直接调用** `skill(name="finishing-a-development-branch")` 和 `skill(name="ship")`
 - 输入: phase-5-summary.md + feedback-log.md → 输出: PR URL
 - **HARD-GATE**: Phase 5 未完成 → BLOCK。验证 `feedback-log.md` 存在。
 - **GITHOOKS-GATE**: 验证 hooks 完整性，缺失则 `githooks/install.sh`
@@ -315,7 +319,7 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 **详细指令**: 参见 `references/phase-7-land.md` — 完整流程、SLA 指标、回滚策略。
 
 **快速参考**:
-- **Dispatch**: `task(category="deep", load_skills=["land-and-deploy"])`
+- **Orchestrator 直接执行**: `land-and-deploy` 包含 merge 确认和 rollback 决策，**必须由 orchestrator 直接调用** `skill(name="land-and-deploy")`
 - 输入: phase-6-summary.md + PR URL → 输出: 部署状态 + Canary 报告
 - 流程: Merge PR → 等待 CI (10min) → 等待 Deploy (10min) → Canary Health Check (5min)
 - **回滚**: `git revert` 最后一次 merge commit

package/skills/sprint-flow/references/orchestration-rules.md

@@ -30,16 +30,32 @@
 |-------|------|:---------:|----------|-------------|--------|
 | -1 | ISOLATE | ❌ | Bash（直接执行） | 无 | orchestrator |
 | -0.5 | AUTO-ESTIMATE | ❌ | Bash（直接执行） | 无 | orchestrator |
-| 0 | THINK | ✅ | `deep` | `["brainstorming"]` | subagent |
-| 1 | PLAN | ✅ | `deep` | `["autoplan", "delphi-review", "to-issues"]` | subagent |
-| 2 | BUILD | ✅(已有) | ralph-loop | `["test-driven-development"]` | subagent |
+| 0 | THINK | ❌ | orchestrator（直接执行） | `["brainstorming"]` | orchestrator |
+| 1 | PLAN | ⚠️ | orchestrator 执行 autoplan + to-issues（交互）→ subagent 执行 delphi-review | 见下方说明 | orchestrator + subagent |
+| 2 | BUILD | ✅ | ralph-loop | `["test-driven-development"]` | subagent |
 | 3 | REVIEW | ✅ | `deep` | `["delphi-review", "test-specification-alignment"]` | subagent |
 | 4 | USER ACCEPT | ❌ | **强制人工** | 无 | 用户 |
 | 5 | FEEDBACK | ✅ | `quick` | `["learn", "retro", "systematic-debugging"]` | subagent |
-| 6 | SHIP | ✅ | `quick` | `["finishing-a-development-branch", "ship"]` | subagent |
-| 7 | LAND | ✅ | `deep` | `["land-and-deploy"]` | subagent |
+| 6 | SHIP | ❌ | orchestrator（直接执行） | `["finishing-a-development-branch", "ship"]` | orchestrator |
+| 7 | LAND | ❌ | orchestrator（直接执行） | `["land-and-deploy"]` | orchestrator |
 | 8 | CLEANUP | ❌ | Bash（直接执行） | 无 | orchestrator |
 
+**⚠️ 交互式 skill 必须由 orchestrator 直接执行（不可 dispatch 到 subagent）**：
+
+| Phase | Skill | 为什么不能在 subagent 中执行 |
+|-------|-------|---------------------------|
+| 0 | `brainstorming` | 需要与用户对话确认需求、提出澄清问题。Subagent 是 fire-and-forget 模式，无法暂停等待用户输入（Issue #217） |
+| 1 | `autoplan` | taste_decisions 节点暂停等待用户确认。必须由 orchestrator 直接执行（Issue #225） |
+| 1 | `to-issues` | Step 6 "向用户确认" — 展示拆分结果，等待用户批准后才生成 slices-manifest.json |
+| 6 | `finishing-a-development-branch` | 4 选项菜单 (merge/PR/keep/discard) 需要用户选择；Option 4 (discard) 要求 typed confirmation |
+| 6 | `ship` | PR 创建前需要用户确认；包含 AskUserQuestion STOP 点 |
+| 7 | `land-and-deploy` | Merge 确认、rollback 决策 — 均为用户交互点 |
+
+**Phase 1 执行模式（两阶段）**：
+1. **Orchestrator 直接执行 autoplan**：`skill(name="autoplan")` → 等待用户确认 taste_decisions
+2. **Orchestrator 直接执行 to-issues**：`skill(name="to-issues")` → 等待用户确认 Issue 拆分
+3. **Subagent 执行 delphi-review**：`task(category="deep", load_skills=["delphi-review"])` — 非交互式，可自动运行至 APPROVED
+
 **上下文隔离原则**：
 - 每个 Subagent 在**独立 session** 中启动，不继承 orchestrator 的对话历史
 - orchestrator session 仅接收 subagent 的最终结果摘要（~13,000 tokens/sprint）
@@ -98,6 +114,20 @@
    - 时机：每个 Phase 完成后的 transition 阶段自动展示，用户无需请求
    - 向后兼容：旧版 `sprint-state.json` 缺少 `phase_history` 时，从 `phase` 字段推断状态
 
+### Background Task Resume Protocol (MANDATORY — Issue #248)
+
+After dispatching background agents with `run_in_background=true` (e.g., Phase -0.5 AUTO-ESTIMATE explore agents, Phase 1 delphi-review subagent):
+
+1. **END YOUR RESPONSE** — do NOT poll `background_output()` before receiving `<system-reminder>`
+2. **On `<system-reminder>` notification**: collect ALL completed results via `background_output(task_id="bg_...")`
+3. **When ALL tasks complete**: immediately resume by:
+   a. Synthesizing results into the current phase assessment
+   b. Executing the phase transition gate
+   c. Continuing to the next phase **WITHOUT waiting for human input**
+4. **If any task failed**: collect partial results, log the failure, continue with available data (do NOT block the pipeline)
+
+**The orchestrator MUST treat task completion notifications as an implicit "continue" signal.** This protocol prevents the "human-in-loop stall" where all background tasks complete but the orchestrator waits indefinitely for a human message (Issue #235, #248).
+
 ### Phase Summary 格式（YAML Frontmatter Schema）
 
 每个 `phase-N-summary.md` 必须包含以下 YAML frontmatter：

package/skills/sprint-flow/references/phase-minus-0-5-auto-estimate.md

@@ -228,6 +228,12 @@ git diff --stat HEAD 2>/dev/null  # 如果有局部修改
 
 ---
 
+## 编排注意事项
+
+Phase -0.5 经常使用并行 background explore agents 评估代码影响范围。Agent dispatch 后必须遵循 **Background Task Resume Protocol**（参见 `orchestration-rules.md`），在所有 background task 返回后自动恢复继续，不得等待人工消息（Issue #248）。
+
+---
+
 ## 与学习循环的集成
 
 ### 数据收集

package/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-06-23
-**Commit:** 35f0341
+**Commit:** c4873e5
 **Branch:** main
-**Version:** 0.10.6.0
+**Version:** 0.10.7.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.