@boyingliu01/xp-gate 0.9.4 → 0.10.0

package/plugins/opencode/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@boyingliu01/opencode-plugin",
-  "version": "0.9.4",
+  "version": "0.10.0",
   "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:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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/delphi-review/SKILL.md

@@ -1,6 +1,7 @@
 ---
 name: delphi-review
 description: "Use when asked to review a design, plan, or architecture; before implementation starts; or when multi-expert consensus is needed. Triggers: 'review this design', '评审这个需求', 'design review', '多专家评审', 'consensus review', 'code walkthrough', 'push review', 'architecture review', 'PR review', or any request for multi-expert evaluation of requirements, design docs, or PRs."
+auto_continue: true
 ---
 
 # Delphi Consensus Review
@@ -109,6 +110,39 @@ Phase 0: 准备 → Round 1: 匿名独立评审 → 共识检查
 
 **Automatic re-review**: 对于常见可控问题（措辞模糊、AC 缺失、格式问题），subagent 应自行修复后自动重评审，无需等待用户。
 
+### ⭐ 自动延续规则（MANDATORY — 防止流程卡住）
+
+当 Delphi Review 启动多轮评审（Round 2+）时，**orchestrator 必须自动延续流程，不得等待用户输入**。
+
+**触发条件**：
+- Round N 完成，但未达到终止条件（100% approved + ≥90% consensus + 所有 Critical/Major 已处理）
+- 存在待处理的背景任务（subagent  dispatched tasks）
+
+**自动延续动作**：
+1. **收集背景任务结果**：等待 `<system-reminder>` 通知后，立即调用 `background_output(task_id="bg_...")` 获取所有 subagent 输出
+2. **合成 Round N 总结**：汇总专家意见、共识度、待处理问题
+3. **自动启动 Round N+1**：立即 dispatch 新一轮 subagent 任务，携带上一轮总结作为上下文
+4. **循环直至终止**：重复步骤 1-3，直到达到终止条件
+
+**终止条件**（满足全部）：
+- ✅ 所有专家状态 = APPROVED
+- ✅ 共识度 ≥ 90%
+- ✅ 所有 Critical 级别问题已解决
+- ✅ 所有 Major 级别问题已处理（或已记录为已知问题）
+
+**例外情况**（直接输出，不进入下一轮）：
+- Round 1 即达到 100% approved + 100% consensus → 直接输出最终报告
+- 已达最大轮数（5 轮）→ 输出"未达成共识"报告，标记为 PROCESS_BLOCK
+
+**禁止行为**：
+- ❌ 询问用户"要继续吗？"
+- ❌ 等待用户手动触发 Round N+1
+- ❌ 在未达到终止条件时停止流程
+
+**错误处理**：
+- 背景任务超时（>10min）→ 标记为 TIMEOUT，输出部分结果并终止
+- 背景任务失败（subagent 错误）→ 重试 1 次，仍失败则输出错误报告并终止
+
 ---
 
 ## 修复与重新评审

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

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
-**Generated:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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

@@ -408,6 +408,58 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 
 ---
 
+### ⭐ Phase State Persistence（阶段状态持久化 — MANDATORY）
+
+**编排器必须在每个 Phase 完成后更新 `.sprint-state/sprint-state.json`**：
+
+1. **Phase 完成后立即更新**（每个 Phase 结束前）：
+   - `phase`: 更新为当前 Phase 编号（如 `0`, `1`, `2`...）
+   - `status`: 更新为 `"completed"`（已完成 Phase）
+   - `phase_history`: 追加新条目
+
+2. **`phase_history` 数组条目 schema**：
+   ```json
+   {
+     "phase": 0,
+     "phase_name": "THINK",
+     "status": "completed",
+     "timestamp": "2026-06-20T10:30:00Z"
+   }
+   ```
+
+3. **检查点**：
+   - `--status` 参数读取 `sprint-state.json` 并渲染进度看板
+   - TUI panel 显示当前 Phase 和历史
+   - `--resume-from` 校验 `phase_history` 中的最后完成 Phase
+
+4. **完整 sprint-state.json 示例**：
+   ```json
+   {
+     "id": "sprint-2026-06-20-01",
+     "phase": 2,
+     "status": "in_progress",
+     "phase_history": [
+       {"phase": -1, "phase_name": "ISOLATE", "status": "completed", "timestamp": "2026-06-20T10:00:00Z"},
+       {"phase": -0.5, "phase_name": "AUTO-ESTIMATE", "status": "completed", "timestamp": "2026-06-20T10:05:00Z"},
+       {"phase": 0, "phase_name": "THINK", "status": "completed", "timestamp": "2026-06-20T10:15:00Z"},
+       {"phase": 1, "phase_name": "PLAN", "status": "completed", "timestamp": "2026-06-20T10:30:00Z"}
+     ],
+     "isolation": {
+       "worktree_path": "/home/boyingliu01/projects/xp-gate/.worktrees/sprint/sprint-2026-06-20-01",
+       "branch": "sprint/2026-06-20-01"
+     },
+     "outputs": {
+       "pain_document": "phase-outputs/phase-0-summary.md",
+       "specification": "phase-outputs/specification.yaml"
+     },
+     "metrics": {
+       "coverage_pct": 85.5
+     }
+   }
+   ```
+
+---
+
 ## 使用示例
 
 | 场景 | 命令 | 说明 |
@@ -447,7 +499,7 @@ See [Output Contract](#output-contract) below for the canonical machine-readable
 
 **Phase Summary** (每个 Phase 必须输出 YAML frontmatter): `phase/N`, `phase_name`, `status`, `outputs[]`, `decisions[]`, `next_phase_context` + markdown body (≤50 lines)
 
-**Sprint State JSON**: `{id, phase, status, isolation {worktree_path, branch}, outputs, metrics}` — 存储于 `.sprint-state/sprint-state.json`
+**Sprint State JSON**: `{id, phase, status, phase_history[], isolation {worktree_path, branch}, outputs, metrics}` — 存储于 `.sprint-state/sprint-state.json`
 
 **Final User-Facing Output**: Phase/status, file paths, validation results, next user decision, PR URL or cleanup report
 

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

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
-**Generated:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/plugins/opencode/tui-plugin.ts

@@ -15,7 +15,67 @@
 import { existsSync, readFileSync } from "node:fs"
 import { join } from "node:path"
 import type { TuiPlugin, TuiSlotPlugin, TuiSlotProps } from "@opencode-ai/plugin/tui"
-import { PHASE_NAMES, PHASE_ORDER, getLatestTimestamp, isStale } from "../../src/npm-package/lib/shared-phase-constants.js"
+// ── Phase constants (inlined from ../../src/npm-package/lib/shared-phase-constants.js)
+//   This file is inlined because the installed npm package does not bundle src/ at publish time. ──
+
+const PHASE_NAMES: Record<string, string> = {
+  '-1': 'ISOLATE',
+  '-0.5': 'AUTO-ESTIMATE',
+  '0': 'THINK',
+  '1': 'PLAN',
+  '2': 'BUILD',
+  '3': 'REVIEW',
+  '4': 'USER ACCEPT',
+  '5': 'FEEDBACK',
+  '6': 'SHIP',
+  '7': 'LAND',
+  '8': 'CLEANUP',
+};
+
+const PHASE_ORDER = ['-1', '-0.5', '0', '1', '2', '3', '4', '5', '6', '7', '8'];
+
+function parseTime(value: unknown): number {
+  return new Date(value as string).getTime();
+}
+
+function maxValid(current: number, candidate: unknown): number {
+  if (!candidate) return current;
+  const t = parseTime(candidate);
+  return !isNaN(t) && t > current ? t : current;
+}
+
+function getLatestTimestamp(state: Record<string, unknown> | null): number {
+  if (!state || !state.started_at) return 0;
+  const started = parseTime(state.started_at);
+  if (isNaN(started)) return 0;
+  let latest = started;
+  if (Array.isArray(state.phase_history)) {
+    for (const ph of state.phase_history) {
+      latest = maxValid(maxValid(latest, (ph as Record<string, unknown>).completed_at), (ph as Record<string, unknown>).started_at);
+    }
+  }
+  return latest;
+}
+
+function isStale(state: SprintState | null): boolean {
+  if (!state || !state.started_at) return false;
+  const latest = sprintTimestamp(state);
+  return latest > 0 && Date.now() - latest > 3600000;
+}
+
+function sprintTimestamp(state: SprintState | null): number {
+  if (!state || !state.started_at) return 0;
+  const started = parseTime(state.started_at);
+  if (isNaN(started)) return 0;
+  let latest = started;
+  if (Array.isArray(state.phase_history)) {
+    for (const ph of state.phase_history) {
+      latest = ph.completed_at ? Math.max(latest, parseTime(ph.completed_at)) : latest;
+      latest = ph.started_at ? Math.max(latest, parseTime(ph.started_at)) : latest;
+    }
+  }
+  return latest;
+}
 
 // ── Sprint state schema ──
 
@@ -101,6 +161,12 @@ function renderBuildReqs(history: SprintPhaseHistory): string[] {
     .map(([id, r]) => `  ${statusSymbol(r.status, id, undefined)} ${r.name}`)
 }
 
+function appendBuildReqs(lines: string[], key: string, history: SprintPhaseHistory | undefined): void {
+  if (key === "2" && history?.reqs) {
+    lines.push(...renderBuildReqs(history))
+  }
+}
+
 function renderPhaseLines(
   historyByPhase: Record<string, SprintPhaseHistory>,
   currentPhase: string | number | undefined,
@@ -110,9 +176,7 @@ function renderPhaseLines(
     const history = historyByPhase[key]
     if (!history && String(currentPhase) !== key) continue
     lines.push(renderPhaseLine(key, history, currentPhase))
-    if (key === "2" && history?.reqs) {
-      lines.push(...renderBuildReqs(history))
-    }
+    appendBuildReqs(lines, key, history)
   }
   return lines
 }
@@ -143,8 +207,7 @@ const tuiPlugin: TuiSlotPlugin = {
       if (!state) return null
       const text = renderSprintSidebar(state)
       if (!text) return null
-      // eslint-disable-next-line @typescript-eslint/no-explicit-any
-      return text as any
+      return text
     },
   },
 }

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

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
-**Generated:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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/delphi-review/SKILL.md

@@ -1,6 +1,7 @@
 ---
 name: delphi-review
 description: "Use when asked to review a design, plan, or architecture; before implementation starts; or when multi-expert consensus is needed. Triggers: 'review this design', '评审这个需求', 'design review', '多专家评审', 'consensus review', 'code walkthrough', 'push review', 'architecture review', 'PR review', or any request for multi-expert evaluation of requirements, design docs, or PRs."
+auto_continue: true
 ---
 
 # Delphi Consensus Review
@@ -109,6 +110,39 @@ Phase 0: 准备 → Round 1: 匿名独立评审 → 共识检查
 
 **Automatic re-review**: 对于常见可控问题（措辞模糊、AC 缺失、格式问题），subagent 应自行修复后自动重评审，无需等待用户。
 
+### ⭐ 自动延续规则（MANDATORY — 防止流程卡住）
+
+当 Delphi Review 启动多轮评审（Round 2+）时，**orchestrator 必须自动延续流程，不得等待用户输入**。
+
+**触发条件**：
+- Round N 完成，但未达到终止条件（100% approved + ≥90% consensus + 所有 Critical/Major 已处理）
+- 存在待处理的背景任务（subagent  dispatched tasks）
+
+**自动延续动作**：
+1. **收集背景任务结果**：等待 `<system-reminder>` 通知后，立即调用 `background_output(task_id="bg_...")` 获取所有 subagent 输出
+2. **合成 Round N 总结**：汇总专家意见、共识度、待处理问题
+3. **自动启动 Round N+1**：立即 dispatch 新一轮 subagent 任务，携带上一轮总结作为上下文
+4. **循环直至终止**：重复步骤 1-3，直到达到终止条件
+
+**终止条件**（满足全部）：
+- ✅ 所有专家状态 = APPROVED
+- ✅ 共识度 ≥ 90%
+- ✅ 所有 Critical 级别问题已解决
+- ✅ 所有 Major 级别问题已处理（或已记录为已知问题）
+
+**例外情况**（直接输出，不进入下一轮）：
+- Round 1 即达到 100% approved + 100% consensus → 直接输出最终报告
+- 已达最大轮数（5 轮）→ 输出"未达成共识"报告，标记为 PROCESS_BLOCK
+
+**禁止行为**：
+- ❌ 询问用户"要继续吗？"
+- ❌ 等待用户手动触发 Round N+1
+- ❌ 在未达到终止条件时停止流程
+
+**错误处理**：
+- 背景任务超时（>10min）→ 标记为 TIMEOUT，输出部分结果并终止
+- 背景任务失败（subagent 错误）→ 重试 1 次，仍失败则输出错误报告并终止
+
 ---
 
 ## 修复与重新评审

package/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
-**Generated:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## 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

@@ -408,6 +408,58 @@ Sprint Flow: ISOLATE → AUTO-ESTIMATE → THINK → PLAN → BUILD → REVIEW
 
 ---
 
+### ⭐ Phase State Persistence（阶段状态持久化 — MANDATORY）
+
+**编排器必须在每个 Phase 完成后更新 `.sprint-state/sprint-state.json`**：
+
+1. **Phase 完成后立即更新**（每个 Phase 结束前）：
+   - `phase`: 更新为当前 Phase 编号（如 `0`, `1`, `2`...）
+   - `status`: 更新为 `"completed"`（已完成 Phase）
+   - `phase_history`: 追加新条目
+
+2. **`phase_history` 数组条目 schema**：
+   ```json
+   {
+     "phase": 0,
+     "phase_name": "THINK",
+     "status": "completed",
+     "timestamp": "2026-06-20T10:30:00Z"
+   }
+   ```
+
+3. **检查点**：
+   - `--status` 参数读取 `sprint-state.json` 并渲染进度看板
+   - TUI panel 显示当前 Phase 和历史
+   - `--resume-from` 校验 `phase_history` 中的最后完成 Phase
+
+4. **完整 sprint-state.json 示例**：
+   ```json
+   {
+     "id": "sprint-2026-06-20-01",
+     "phase": 2,
+     "status": "in_progress",
+     "phase_history": [
+       {"phase": -1, "phase_name": "ISOLATE", "status": "completed", "timestamp": "2026-06-20T10:00:00Z"},
+       {"phase": -0.5, "phase_name": "AUTO-ESTIMATE", "status": "completed", "timestamp": "2026-06-20T10:05:00Z"},
+       {"phase": 0, "phase_name": "THINK", "status": "completed", "timestamp": "2026-06-20T10:15:00Z"},
+       {"phase": 1, "phase_name": "PLAN", "status": "completed", "timestamp": "2026-06-20T10:30:00Z"}
+     ],
+     "isolation": {
+       "worktree_path": "/home/boyingliu01/projects/xp-gate/.worktrees/sprint/sprint-2026-06-20-01",
+       "branch": "sprint/2026-06-20-01"
+     },
+     "outputs": {
+       "pain_document": "phase-outputs/phase-0-summary.md",
+       "specification": "phase-outputs/specification.yaml"
+     },
+     "metrics": {
+       "coverage_pct": 85.5
+     }
+   }
+   ```
+
+---
+
 ## 使用示例
 
 | 场景 | 命令 | 说明 |
@@ -447,7 +499,7 @@ See [Output Contract](#output-contract) below for the canonical machine-readable
 
 **Phase Summary** (每个 Phase 必须输出 YAML frontmatter): `phase/N`, `phase_name`, `status`, `outputs[]`, `decisions[]`, `next_phase_context` + markdown body (≤50 lines)
 
-**Sprint State JSON**: `{id, phase, status, isolation {worktree_path, branch}, outputs, metrics}` — 存储于 `.sprint-state/sprint-state.json`
+**Sprint State JSON**: `{id, phase, status, phase_history[], isolation {worktree_path, branch}, outputs, metrics}` — 存储于 `.sprint-state/sprint-state.json`
 
 **Final User-Facing Output**: Phase/status, file paths, validation results, next user decision, PR URL or cleanup report
 

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

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
-**Generated:** 2026-06-19
-**Commit:** b67f7f9
+**Generated:** 2026-06-22
+**Commit:** 984d568
 **Branch:** main
-**Version:** 0.9.4.0
+**Version:** 0.10.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.