iterate-plugin 3.16.0 → 3.17.0

package/README.md

@@ -107,7 +107,7 @@ Both modes share:
 ### UI layer (build-free client slots, v3.12: 12 tabs, bilingual)
 
 - **ConvergenceDashboard** (`conversation.input.dock`) — live round progress bar, severity stats, dimension badges, trend mini-chart above the input; normal mode also shows fix-count badges; plus a live workflow-phase chip (current phase + running/stopped) and a **task_mode indicator (code / iterate)**, now backed by the persisted transcript (`iterate_status` / `iterate_transcript` emit `taskMode`). While a run is active the dashboard switches to a running state (round + phase) instead of the static empty state, and the start buttons are hidden.
-- **ObservatoryPanel** (`conversation.input.dock`) — a **twelve-tab** runtime observatory below the input: **live activity stream** (default tab, type filter) + **F1** review threads (expand / collapse all), **F2** convergence trend (incl. the per-round comparison table: findings raised / fixed / severity mix / validation outcomes, plus **v3.16 a cross-session comparison**: save the current run as a baseline (`存为对比基准`), then compare any later run against it side by side (`与基准对比`) with per-metric deltas tinted by direction), **F3** finding locations (severity / dimension / search filter), **F4** fixes + rollback, **F5** checkpoint resume, **F6** run console, **F7** decision timeline (type / round filter + search), **F8** Quality Gate, **F9** Experience Bank, **F10** Defense Events, **F11** Preflight. Number-key shortcuts: `1`–`9` jump to F1–F9, `0` to F10, `-` back to the live stream (F11 is click-only). **F8/F9/F10 are live** — they scan the session stream for the latest `iterate_quality_gate` / `iterate_experience` / `iterate_defense_events` results and render real data (PASS/FAIL chip + per-dimension score bars, experience entries with adopt buttons + client-side search, defense type-count chips + filtered event stream); when the session stream has none they distinguish "no data on disk" from "not pulled yet" and offer a **拉取磁盘快照 (pull disk snapshot)** instruction that has the model call the read-side tools and echo the `.iterate/` artifacts back into the stream. One-click JSON export of all observatory data — manifest + live + `qualityGate` / `experienceBank` / `defenseEvents` / `report` (download, copy fallback). **v3.9: report-export buttons** (导出报告 / 导出 SARIF, **v3.11 + 导出 JSON 报告**) render the *same shared data model* as the `iterate_report` tool — a PR-pasteable Markdown run report, a SARIF 2.1.0 log for CI / GitHub Code Scanning, and (v3.11) the complete normalized run model as machine-readable JSON. **v3.5.3: reset actions surfaced** — F5 copies an `iterate_checkpoint clear` instruction, F8 an `iterate_quality_gate clear` instruction, F10 an `iterate_defense_events clear` instruction. **v3.12: pre-run readiness (Preflight)** — F11 renders the latest `iterate_preflight` checklist scanned from the session stream (validation commands configured / git repo / dirty tree / stale FAIL gate / interrupted checkpoint), each warn/fail row offering the exact fix instruction (configure `validation.commands`, clear the stale gate, resume the checkpoint) so the environment is repaired BEFORE a round is spent on it; the empty state offers a **运行预检** instruction that has the model call `iterate_preflight`. **v3.13: the blocking validation row auto-detects commands from the project manifest and offers a one-click 「应用建议」 (copies an `iterate_config write` payload).**
+- **ObservatoryPanel** (`conversation.input.dock`) — a **twelve-tab** runtime observatory below the input: **live activity stream** (default tab, type filter) + **F1** review threads (expand / collapse all), **F2** convergence trend (incl. the per-round comparison table: findings raised / fixed / severity mix / validation outcomes, plus **v3.16 a cross-session comparison**: save the current run as a baseline (`存为对比基准`), then compare any later run against it side by side (`与基准对比`) with per-metric deltas tinted by direction, **v3.17 the comparison is copyable as a Markdown table** (`复制对比表`, paste it straight into a PR / issue / CI comment)), **F3** finding locations (severity / dimension / search filter), **F4** fixes + rollback, **F5** checkpoint resume, **F6** run console, **F7** decision timeline (type / round filter + search), **F8** Quality Gate (query / **v3.17 one-click recompute** — the `compute` payload's per-dimension per-round convergence series is derived from the session transcript, so you no longer hand-write the `findingsByRound` JSON), **F9** Experience Bank, **F10** Defense Events, **F11** Preflight. Number-key shortcuts: `1`–`9` jump to F1–F9, `0` to F10, `-` back to the live stream (F11 is click-only). **F8/F9/F10 are live** — they scan the session stream for the latest `iterate_quality_gate` / `iterate_experience` / `iterate_defense_events` results and render real data (PASS/FAIL chip + per-dimension score bars, experience entries with adopt buttons + client-side search, defense type-count chips + filtered event stream); when the session stream has none they distinguish "no data on disk" from "not pulled yet" and offer a **拉取磁盘快照 (pull disk snapshot)** instruction that has the model call the read-side tools and echo the `.iterate/` artifacts back into the stream. One-click JSON export of all observatory data — manifest + live + `qualityGate` / `experienceBank` / `defenseEvents` / `report` (download, copy fallback). **v3.9: report-export buttons** (导出报告 / 导出 SARIF, **v3.11 + 导出 JSON 报告**) render the *same shared data model* as the `iterate_report` tool — a PR-pasteable Markdown run report, a SARIF 2.1.0 log for CI / GitHub Code Scanning, and (v3.11) the complete normalized run model as machine-readable JSON. **v3.5.3: reset actions surfaced** — F5 copies an `iterate_checkpoint clear` instruction, F8 an `iterate_quality_gate clear` instruction, F10 an `iterate_defense_events clear` instruction. **v3.12: pre-run readiness (Preflight)** — F11 renders the latest `iterate_preflight` checklist scanned from the session stream (validation commands configured / git repo / dirty tree / stale FAIL gate / interrupted checkpoint), each warn/fail row offering the exact fix instruction (configure `validation.commands`, clear the stale gate, resume the checkpoint) so the environment is repaired BEFORE a round is spent on it; the empty state offers a **运行预检** instruction that has the model call `iterate_preflight`. **v3.13: the blocking validation row auto-detects commands from the project manifest and offers a one-click 「应用建议」 (copies an `iterate_config write` payload).**
 - **TriagePanel** (`conversation.chat.turnTail`) — per-finding triage with `y` (keep → fix), `n` (skip), `a` (known-intentional) keyboard shortcuts and `↑`/`↓` navigation (focus-guarded: only active while the panel has focus), filtering, batch ops (incl. select-all), localStorage persistence. **Write-back closed loop**: the primary button copies an `iterate_triage apply` instruction for the marked entries (the tool validates, dedupes, backs up, rolls back on failure); a secondary button copies the raw `known_intentional` YAML; **回读已写入条目** copies an `iterate_triage list` instruction to confirm what actually landed. Copy payloads are aligned to the target tool schemas. **v3.2: native command buttons** (approve architectural fix, trigger new round, rollback to checkpoint). **v3.3: §8 指派修复 (assign)** — one click copies an `iterate_fix` instruction carrying every in-scope finding (respecting the batch toggle).
 - **StatsCard** (`conversation.chat.turnTail`) — convergence stats, round history, trend chart and completion summary when no findings remain.
 - **iterate theme skin** (`theme.overrideTokens`) — a warm-amber 13-token `--dsw-*` override, light/dark modes, togglable in settings.
@@ -121,7 +121,7 @@ The UI layer is **defensive by design**: it degrades gracefully if any of `slots
 
 ### Behavior beyond the tools
 
-Beyond the 17 registered tools (full reference further down), the plugin closes several loops end to end:
+Beyond the 19 registered tools (full reference further down), the plugin closes several loops end to end:
 
 - **Findings triage loop** — review → UI triage (y/n/a) → `iterate_triage` writes back `known_intentional` → auto-filtered next round
 - **Structured fix system** — each fix backs up first, writes a registry entry, records the diff; a failed validation can be reverted with `iterate_rollback`

package/README.zh-CN.md

@@ -106,7 +106,7 @@ dsh plugin --profile web add iterate-plugin
 ### UI 层（客户端免构建槽位，v3.12：12 个页签，双语）
 
 - **收敛看板 `ConvergenceDashboard`**（`conversation.input.dock`）— 输入框上方实时显示轮次进度条、严重度统计、维度徽章、趋势迷你图；normal 模式另显示修复计数徽章；还有运行阶段芯片（当前工作流阶段 + 运行中/已结束）与 **task_mode 指示器（code / iterate）**。运行期间看板切换为"运行中"状态（显示轮次与阶段），并隐藏启动按钮。
-- **运行时观测台 `ObservatoryPanel`**（`conversation.input.dock`）— 输入框下方**十二个页签**运行时观测台：**实时活动流**（默认页，支持按活动类型筛选）+ **F1** 审查线程（全部展开/全部收起）、**F2** 收敛趋势（含跨轮对比表：每轮 findings / fixed / 严重度分布 / 验证结果；**v3.16 新增跨会话对比**：把当前运行**存为对比基准**，下次运行一键**与基准对比**，逐项并排 + 带方向的差值着色）、**F3** 发现定位（按严重度/维度/关键词筛选）、**F4** 修复与回滚、**F5** 断点恢复、**F6** 运行控制台、**F7** 决策时间线（按类型/轮次筛选与关键词搜索）、**F8** 质量门禁、**F9** 经验银行、**F10** 防御事件、**F11** 开跑预检。数字键快捷键：`1`–`9` 跳到 F1–F9，`0` 到 F10，`-` 回实时活动流（F11 仅点击进入）。**F8/F9/F10 为实时数据**——扫描会话流中最近一次 `iterate_quality_gate` / `iterate_experience` / `iterate_defense_events` 结果并渲染真实数据（PASS/FAIL 徽章 + 分维度评分条、经验条目（含采纳按钮 + 客户端搜索）、防御类型计数芯片 + 筛选后的事件流）；会话流中没有数据时区分"磁盘无数据"与"尚未拉取"，并提供**拉取磁盘快照**指令（让模型调用只读工具、把 `.iterate/` 产物回显进会话流）。一键导出全部观测数据为 JSON——manifest + live + `qualityGate` / `experienceBank` / `defenseEvents` / `report`（优先下载，失败回退复制）。**v3.9 新增报告导出按钮**（导出报告 / 导出 SARIF，**v3.11 新增导出 JSON 报告**），与 `iterate_report` 工具共用同一份数据模型——分别产出可直接粘贴进 PR 的 Markdown 运行报告、供 CI / GitHub Code Scanning 消费的 SARIF 2.1.0 日志，以及（v3.11）完整的归一化运行模型 JSON（无损格式）。**v3.5.3 新增重置操作**：F5 复制 `iterate_checkpoint clear`、F8 复制 `iterate_quality_gate clear`、F10 复制 `iterate_defense_events clear` 指令。**v3.12 新增开跑预检（Preflight）**——F11 渲染会话流中最近一次 `iterate_preflight` 清单（验证命令是否配置 / 是否 Git 仓库 / 脏工作区 / 陈旧 FAIL 门禁 / 中断断点），每个 warn/fail 行都给出精确的修复指令（配置 `validation.commands`、清除陈旧门禁、恢复断点），让环境在**花掉一轮之前**就被修好；空态提供**运行预检**指令（让模型调用 `iterate_preflight`）。**v3.13：阻断的验证命令行自动检测项目 manifest，提供一键「应用建议」**（复制 `iterate_config` 的 `operation:"write"` 指令，携带完整 `validation.commands.default` 载荷）。
+- **运行时观测台 `ObservatoryPanel`**（`conversation.input.dock`）— 输入框下方**十二个页签**运行时观测台：**实时活动流**（默认页，支持按活动类型筛选）+ **F1** 审查线程（全部展开/全部收起）、**F2** 收敛趋势（含跨轮对比表：每轮 findings / fixed / 严重度分布 / 验证结果；**v3.16 新增跨会话对比**：把当前运行**存为对比基准**，下次运行一键**与基准对比**，逐项并排 + 带方向的差值着色，**v3.17 新增「复制对比表」**：把并排表复制成 Markdown 表格，可直接粘贴进 PR / issue / CI 注释）、**F3** 发现定位（按严重度/维度/关键词筛选）、**F4** 修复与回滚、**F5** 断点恢复、**F6** 运行控制台、**F7** 决策时间线（按类型/轮次筛选与关键词搜索）、**F8** 质量门禁（查询 / **v3.17 一键重算**——`compute` 载荷里每维度每轮的收敛序列直接从当前会话 transcript 推导，无需再手写 `findingsByRound` JSON）、**F9** 经验银行、**F10** 防御事件、**F11** 开跑预检。数字键快捷键：`1`–`9` 跳到 F1–F9，`0` 到 F10，`-` 回实时活动流（F11 仅点击进入）。**F8/F9/F10 为实时数据**——扫描会话流中最近一次 `iterate_quality_gate` / `iterate_experience` / `iterate_defense_events` 结果并渲染真实数据（PASS/FAIL 徽章 + 分维度评分条、经验条目（含采纳按钮 + 客户端搜索）、防御类型计数芯片 + 筛选后的事件流）；会话流中没有数据时区分"磁盘无数据"与"尚未拉取"，并提供**拉取磁盘快照**指令（让模型调用只读工具、把 `.iterate/` 产物回显进会话流）。一键导出全部观测数据为 JSON——manifest + live + `qualityGate` / `experienceBank` / `defenseEvents` / `report`（优先下载，失败回退复制）。**v3.9 新增报告导出按钮**（导出报告 / 导出 SARIF，**v3.11 新增导出 JSON 报告**），与 `iterate_report` 工具共用同一份数据模型——分别产出可直接粘贴进 PR 的 Markdown 运行报告、供 CI / GitHub Code Scanning 消费的 SARIF 2.1.0 日志，以及（v3.11）完整的归一化运行模型 JSON（无损格式）。**v3.5.3 新增重置操作**：F5 复制 `iterate_checkpoint clear`、F8 复制 `iterate_quality_gate clear`、F10 复制 `iterate_defense_events clear` 指令。**v3.12 新增开跑预检（Preflight）**——F11 渲染会话流中最近一次 `iterate_preflight` 清单（验证命令是否配置 / 是否 Git 仓库 / 脏工作区 / 陈旧 FAIL 门禁 / 中断断点），每个 warn/fail 行都给出精确的修复指令（配置 `validation.commands`、清除陈旧门禁、恢复断点），让环境在**花掉一轮之前**就被修好；空态提供**运行预检**指令（让模型调用 `iterate_preflight`）。**v3.13：阻断的验证命令行自动检测项目 manifest，提供一键「应用建议」**（复制 `iterate_config` 的 `operation:"write"` 指令，携带完整 `validation.commands.default` 载荷）。
 - **Findings 分诊面板 `TriagePanel`**（`conversation.chat.turnTail`）— 逐条分诊，键盘快捷键 `y`（保留→修复）、`n`（跳过）、`a`（已知有意）与 `↑`/`↓` 移动（焦点守卫：仅面板内焦点时生效），支持筛选、批量（含一键全选所有 findings）、localStorage 持久化。**写回闭环**：主按钮复制已标记条目的 `iterate_triage apply` 指令（工具负责校验、去重、备份、失败回滚）；次按钮复制裸 `known_intentional` YAML；**回读已写入条目**复制 `iterate_triage list` 指令核对实际落盘内容。复制载荷与目标工具 schema 对齐。**v3.2：原生命令按钮**（批准架构修复、触发新一轮、回滚到断点）。
 - **收敛统计卡片 `StatsCard`**（`conversation.chat.turnTail`）— 无 findings 时显示收敛统计、历史轮次、趋势图、完成摘要。
 - **iterate 主题皮肤**（`theme.overrideTokens`）— 暖琥珀配色的 13 个 `--dsw-*` token 覆盖，明暗双模式，可在设置页开关。

package/dist/index.js

@@ -2,16 +2,17 @@
  * iterate-plugin — dsh plugin for the iterate autonomous closed-loop workflow
  *
  * Architecture:
- * - The plugin registers 18 tools (14 original + 3 v3.0 quality command center
- *   tools + 1 v3.9 report export)
+ * - The plugin registers 19 tools (14 original + 3 v3.0 quality command center
+ *   tools + 1 v3.9 report export + 1 v3.12 pre-run preflight)
  *   Original: config, validate, decision-log, context, review, triage, fix, diff,
  *   rollback, checkpoint, status, history, prune, transcript
  *   v3.0: experience, quality_gate, defense_events
  *   v3.9: report
+ *   v3.12: preflight
  * - The plugin injects a system prompt section teaching the iterate workflow pattern
  * - The model (prompted by the skill) writes a workflow script using dsh's `workflow` tool
  * - The workflow script uses `agent()` / `parallel()` / `phase()` / `log()` to orchestrate
- * - Subagents use the 18 tools to do real work (read config, run validation, log decisions,
+ * - Subagents use the 19 tools to do real work (read config, run validation, log decisions,
  *   review, triage, apply/rollback/fixing, checkpoint, status, history, prune, transcript,
  *   query experience bank, check quality gates, query defense events, export a report)
  * - A `tools/pre-execute` hook gates destructive iterate calls behind human approval
@@ -25,7 +26,7 @@
  *
  * Key files:
  * - src/index.ts      — Plugin entry: register tools + inject skill prompt
- * - src/tools/        — 18 tool implementations (14 original + 3 v3.0 + 1 v3.9)
+ * - src/tools/        — 19 tool implementations (14 original + 3 v3.0 + 1 v3.9 + 1 v3.12)
  * - src/config-loader.ts — YAML config loading
  * - src/types.ts     — Shared types
  */

package/dist/tools/decision-log.js

@@ -265,6 +265,26 @@ function decodedLines(filePath) {
     }
     return out;
 }
+/**
+ * Rebuild the exact ledger text the kernel verifies, dropping only the artifact a
+ * trailing newline leaves. Unlike `decodedLines` this keeps interior blank lines:
+ * `kernelVerifyLedgerText` splits on "\n" and `JSON.parse`s each line, so an interior
+ * blank line is a corrupt entry it must report — renumbering the lines that follow it
+ * (by dropping it here) would mask a real break as a healthy ledger.
+ */
+function ledgerTextForVerify(filePath) {
+    let text;
+    try {
+        text = readFileSync(filePath, 'utf-8');
+    }
+    catch {
+        return '';
+    }
+    const lines = text.endsWith('\n') ? text.slice(0, -1) : text;
+    if (lines.length === 0)
+        return '';
+    return `${lines}\n`;
+}
 /** The 1-based line number of the first legacy record AFTER the chain began, else `null`. */
 export function contestedLedgerLine(filePath) {
     const kinds = decodedLines(filePath).map((l) => classifyDecisionLine(l.decoded));
@@ -670,8 +690,7 @@ export function readDecisionEntries(projectRoot) {
 export function verifyDecisionLedger(projectRoot) {
     const filePath = decisionLogPath(projectRoot);
     try {
-        const lines = decodedLines(filePath);
-        const body = `${lines.map((l) => l.raw).join('\n')}${lines.length > 0 ? '\n' : ''}`;
+        const body = ledgerTextForVerify(filePath);
         const verified = kernelVerifyLedgerText(body);
         return {
             ok: verified.ok,

package/lib/client.js

@@ -1,4 +1,4 @@
-/* iterate-plugin client source-digest: 55b24ad8a8a5d8abf400d42314a6bf75394e8e529fc587ac991bb4e50d868393 */
+/* iterate-plugin client source-digest: 1396cad72b6ae5e0e09c4fac31f0e74b0f17719bb8a1d3c2c30cc749e34b70c1 */
 window.__ModuleLoader__.load({ id: "iterate-plugin", factory: (require) => {
 var module = { exports: {} };
 var exports = module.exports;
@@ -1343,6 +1343,125 @@ function buildQualityGateClearInstruction() {
 ${JSON.stringify({ operation: "clear" }, null, 2)}
 \`\`\``;
 }
+function buildQualityGateComputeInstruction(input) {
+  const src = input && typeof input === "object" ? input : {};
+  const cleanStr = (v) => typeof v === "string" && v.length > 0;
+  const dimensions = Array.isArray(src.dimensions) ? src.dimensions.filter((d) => typeof d === "string" && d.length > 0) : [];
+  const findings = Array.isArray(src.findings) ? src.findings.filter((f) => f && typeof f === "object" && cleanStr(f.dimension) && cleanStr(f.file)).map((f) => ({
+    dimension: f.dimension,
+    severity: typeof f.severity === "string" && f.severity ? f.severity : "medium",
+    file: f.file,
+    ...typeof f.line === "number" && Number.isFinite(f.line) && f.line > 0 ? { line: f.line } : {}
+  })) : [];
+  const validationResults = Array.isArray(src.validationResults) ? src.validationResults.filter((v) => v && typeof v === "object" && cleanStr(v.command) && typeof v.exitCode === "number" && Number.isFinite(v.exitCode)).map((v) => ({ command: v.command, exitCode: v.exitCode })) : [];
+  const findingsByRound = {};
+  if (src.findingsByRound && typeof src.findingsByRound === "object" && !Array.isArray(src.findingsByRound)) {
+    for (const [dim, series] of Object.entries(src.findingsByRound)) {
+      const clean = Array.isArray(series) ? series.filter((n) => typeof n === "number" && Number.isFinite(n) && n >= 0) : [];
+      if (clean.length > 0) findingsByRound[dim] = clean;
+    }
+  }
+  const fixedByDimension = {};
+  if (src.fixedByDimension && typeof src.fixedByDimension === "object" && !Array.isArray(src.fixedByDimension)) {
+    for (const [dim, n] of Object.entries(src.fixedByDimension)) {
+      if (typeof n === "number" && Number.isFinite(n) && n >= 0) fixedByDimension[dim] = n;
+    }
+  }
+  const payload = { operation: "compute" };
+  if (dimensions.length > 0) payload.dimensions = dimensions;
+  if (findings.length > 0) payload.findings = findings;
+  if (validationResults.length > 0) payload.validationResults = validationResults;
+  if (Object.keys(findingsByRound).length > 0) payload.findingsByRound = findingsByRound;
+  if (Object.keys(fixedByDimension).length > 0) payload.fixedByDimension = fixedByDimension;
+  return `\u8BF7\u8C03\u7528 \`iterate_quality_gate\` \u91CD\u7B97\u8D28\u91CF\u95E8\u7981\u8BC1\u4E66\uFF08\u4ECE\u5F53\u524D\u4F1A\u8BDD\u63A8\u5BFC\u7EF4\u5EA6/\u53D1\u73B0/\u9A8C\u8BC1\u7ED3\u679C\u5E76\u6301\u4E45\u5316\uFF09\uFF1A
+
+\`\`\`json
+${JSON.stringify(payload, null, 2)}
+\`\`\``;
+}
+function deriveQualityGateComputeInput(manifest, qualityGate) {
+  const m = manifest && typeof manifest === "object" ? manifest : {};
+  const asArray2 = (v) => Array.isArray(v) ? v : [];
+  const dims = [];
+  for (const r of asArray2(m.rounds)) {
+    for (const t2 of asArray2(r && r.threads)) {
+      const d = t2 && typeof t2.dimension === "string" ? t2.dimension : "";
+      if (d && !dims.includes(d)) dims.push(d);
+    }
+  }
+  if (dims.length === 0 && qualityGate && Array.isArray(qualityGate.dimensions)) {
+    for (const d of qualityGate.dimensions) {
+      const name2 = d && typeof d.dimension === "string" ? d.dimension : "";
+      if (name2 && !dims.includes(name2)) dims.push(name2);
+    }
+  }
+  const findings = [];
+  for (const f of asArray2(m.findings)) {
+    if (!f || typeof f !== "object") continue;
+    if (typeof f.dimension !== "string" || !f.dimension) continue;
+    if (typeof f.file !== "string" || !f.file) continue;
+    findings.push({
+      dimension: f.dimension,
+      severity: typeof f.severity === "string" && f.severity ? f.severity : "medium",
+      file: f.file,
+      ...typeof f.line === "number" && Number.isFinite(f.line) && f.line > 0 ? { line: f.line } : {}
+    });
+  }
+  const validationResults = [];
+  for (const v of asArray2(m.validations)) {
+    if (!v || typeof v !== "object") continue;
+    if (typeof v.command !== "string" || !v.command) continue;
+    if (typeof v.exitCode !== "number" || !Number.isFinite(v.exitCode)) continue;
+    validationResults.push({ command: v.command, exitCode: v.exitCode });
+  }
+  const rounds = asArray2(m.rounds);
+  const allDims = dims.slice();
+  for (const r of rounds) {
+    for (const t2 of asArray2(r && r.threads)) {
+      const d = t2 && typeof t2.dimension === "string" ? t2.dimension : "";
+      if (d && !allDims.includes(d)) allDims.push(d);
+    }
+  }
+  const byRound = {};
+  for (const d of allDims) {
+    byRound[d] = [];
+  }
+  for (const r of rounds) {
+    for (const d of allDims) {
+      let n = 0;
+      for (const t2 of asArray2(r && r.threads)) {
+        if (!t2 || t2.dimension !== d) continue;
+        for (const f of asArray2(t2 && t2.findings)) {
+          if (f && typeof f === "object" && typeof f.dimension === "string" && f.dimension === d) n += 1;
+        }
+      }
+      byRound[d].push(n);
+    }
+  }
+  const fixedByDimension = {};
+  const findingsIndex = findings;
+  for (const fx of asArray2(m.fixes)) {
+    if (!fx || typeof fx !== "object" || fx.success !== true) continue;
+    const file = typeof fx.file === "string" ? fx.file : "";
+    if (!file) continue;
+    let dim = "";
+    for (const f of findingsIndex) {
+      if (f.file !== file) continue;
+      if (typeof f.line === "number" && typeof fx.line === "number" && f.line !== fx.line) continue;
+      dim = f.dimension;
+      break;
+    }
+    if (!dim) continue;
+    fixedByDimension[dim] = (fixedByDimension[dim] || 0) + 1;
+  }
+  return {
+    dimensions: dims,
+    findings,
+    validationResults,
+    findingsByRound: byRound,
+    fixedByDimension
+  };
+}
 function buildPreflightInstruction() {
   return `\u8BF7\u8C03\u7528 \`iterate_preflight\` \u8FD0\u884C\u5F00\u8DD1\u524D\u9884\u68C0\uFF08\u53EA\u8BFB\uFF1A\u6821\u9A8C validation.commands\u3001git\u3001\u810F\u5DE5\u4F5C\u533A\u3001\u9648\u65E7\u8D28\u91CF\u95E8\u7981\u4E0E\u4E2D\u65AD\u65AD\u70B9\uFF09\uFF1A
 
@@ -2663,6 +2782,8 @@ var EN = {
   \u57FA\u51C6: "Baseline",
   \u5F53\u524D: "Current",
   \u53D8\u5316: "Delta",
+  \u590D\u5236\u5BF9\u6BD4\u8868: "Copy comparison table",
+  "\u590D\u5236\u8DE8\u4F1A\u8BDD\u5BF9\u6BD4\u8868\u4E3A Markdown\uFF08\u53EF\u7C98\u8D34\u8FDB PR / issue / CI \u6CE8\u91CA\uFF09": "Copy the cross-session comparison as a Markdown table (paste into a PR / issue / CI comment)",
   // ── F3 finding locations ─────────────────────────────────────────────────
   \u6682\u65E0\u53D1\u73B0: "No findings",
   "{a0}\uFF08{a1} \u9879\uFF09": "{a0} ({a1} findings)",
@@ -2729,6 +2850,9 @@ var EN = {
   \u67E5\u8BE2\u95E8\u7981: "Query gate",
   "\u590D\u5236 iterate_quality_gate clear \u6307\u4EE4\u6587\u672C\uFF08\u91CD\u7F6E\u9648\u65E7\u7684 FAIL \u8BC1\u4E66\uFF09": "Copy the iterate_quality_gate clear instruction (resets a stale FAIL certificate)",
   \u6E05\u9664\u95E8\u7981: "Clear gate",
+  \u91CD\u7B97\u95E8\u7981: "Recompute gate",
+  "\u590D\u5236 iterate_quality_gate compute \u6307\u4EE4\uFF1A\u4ECE\u5F53\u524D\u4F1A\u8BDD\u63A8\u5BFC\u7EF4\u5EA6/\u53D1\u73B0/\u9A8C\u8BC1\u7ED3\u679C\uFF0C\u4E00\u952E\u91CD\u7B97\u5E76\u6301\u4E45\u5316\u95E8\u7981\u8BC1\u4E66": "Copy the iterate_quality_gate compute instruction: derive dimensions / findings / validation results from this session and recompute + persist the gate certificate",
+  "\u5F53\u524D\u4F1A\u8BDD\u6CA1\u6709\u53EF\u91CD\u7B97\u7684\u95E8\u7981\u6570\u636E\uFF08\u9700\u5148\u6709\u5BA1\u67E5/\u9A8C\u8BC1\u7ED3\u679C\uFF09": "No gate data in this session to recompute (need review / validation results first)",
   "\u590D\u5236\u62C9\u53D6\u78C1\u76D8\u5FEB\u7167\u6307\u4EE4\uFF1A\u8BA9\u6A21\u578B\u56DE\u663E .iterate/ \u4E0B\u7684\u95E8\u7981\u3001\u7ECF\u9A8C\u3001\u9632\u5FA1\u4E8B\u4EF6\u4E0E transcript": "Copy the disk-snapshot instruction: have the model echo the gate, experience, defense events and transcript artifacts under .iterate/",
   \u8D28\u91CF\u95E8\u7981\u89C6\u56FE: "Quality gate view",
   \u8D28\u91CF\u95E8\u7981\u8BC1\u4E66: "Quality gate certificate",
@@ -3153,6 +3277,21 @@ function compareRuns(baseline, current3) {
   ];
   return rows;
 }
+function buildRunCompareMarkdown(rows, labels) {
+  const cell = (v) => v === null ? "\u2014" : String(v);
+  const header = ["\u6307\u6807", "\u57FA\u51C6", "\u5F53\u524D", "\u53D8\u5316"].map((h) => labels[h] ?? h);
+  const lines = [
+    `| ${header.join(" | ")} |`,
+    `| ${header.map(() => "---").join(" | ")} |`
+  ];
+  for (const row of rows) {
+    const delta = row.baseline === null || row.current === null ? "\u2014" : String(row.current - row.baseline);
+    lines.push(
+      `| ${labels[row.key] ?? row.key} | ${cell(row.baseline)} | ${cell(row.current)} | ${delta} |`
+    );
+  }
+  return lines.join("\n");
+}
 
 // src/client/index.ts
 var React = require("react");
@@ -5375,6 +5514,20 @@ function ObservatoryPanel(props) {
             title: t("\u663E\u793A\u6216\u9690\u85CF\u4E0E\u5DF2\u4FDD\u5B58\u57FA\u51C6\u7684\u5BF9\u6BD4\u8868"),
             onClick: () => setShowCompare((v) => !v)
           }, showCompare ? t("\u6536\u8D77\u5BF9\u6BD4") : t("\u4E0E\u57FA\u51C6\u5BF9\u6BD4")) : null,
+          baseline && current3 && showCompare ? React.createElement("button", {
+            className: "iterate-btn",
+            "data-copied": copiedKey === "rc-copy" ? "" : void 0,
+            onClick: () => copyInstruction("rc-copy", buildRunCompareMarkdown(compareRuns(baseline, current3), {
+              \u6307\u6807: t("\u6307\u6807"),
+              \u57FA\u51C6: t("\u57FA\u51C6"),
+              \u5F53\u524D: t("\u5F53\u524D"),
+              \u53D8\u5316: t("\u53D8\u5316"),
+              ...Object.fromEntries(
+                Object.entries(RUN_COMPARE_METRIC_LABELS).map(([k, v]) => [k, v[getLang()]])
+              )
+            })),
+            title: t("\u590D\u5236\u8DE8\u4F1A\u8BDD\u5BF9\u6BD4\u8868\u4E3A Markdown\uFF08\u53EF\u7C98\u8D34\u8FDB PR / issue / CI \u6CE8\u91CA\uFF09")
+          }, copiedKey === "rc-copy" ? t("\u5DF2\u590D\u5236") : t("\u590D\u5236\u5BF9\u6BD4\u8868")) : null,
           baseline ? React.createElement("button", {
             className: "iterate-btn",
             "data-ghost": "",
@@ -5820,6 +5973,12 @@ function ObservatoryPanel(props) {
   const renderQualityGate = () => {
     const gateInstruction = buildQualityGateQueryInstruction();
     const gateClearInstruction = buildQualityGateClearInstruction();
+    const computeInput = deriveQualityGateComputeInput(
+      manifest,
+      qualityGate
+    );
+    const computeInstruction = buildQualityGateComputeInstruction(computeInput);
+    const hasComputeData = computeInput.dimensions.length > 0 || computeInput.findings.length > 0 || computeInput.validationResults.length > 0;
     const gate = qualityGate;
     const dims = gate && gate.dimensions ? gate.dimensions : [];
     const hasGate = gate !== null && Boolean(gate.overallStatus || gate.overallScore != null || dims.length > 0);
@@ -5838,6 +5997,14 @@ function ObservatoryPanel(props) {
         onClick: () => copyInstruction("qgate", gateInstruction),
         title: t("\u590D\u5236 iterate_quality_gate \u67E5\u8BE2\u6307\u4EE4\uFF08\u542B JSON \u53C2\u6570\uFF09")
       }, copiedKey === "qgate" ? t("\u5DF2\u590D\u5236") : t("\u67E5\u8BE2\u95E8\u7981")),
+      React.createElement("button", {
+        className: "iterate-btn",
+        "data-primary": "",
+        "data-copied": copiedKey === "qgate-compute" ? "" : void 0,
+        disabled: !hasComputeData,
+        onClick: () => copyInstruction("qgate-compute", computeInstruction),
+        title: hasComputeData ? t("\u590D\u5236 iterate_quality_gate compute \u6307\u4EE4\uFF1A\u4ECE\u5F53\u524D\u4F1A\u8BDD\u63A8\u5BFC\u7EF4\u5EA6/\u53D1\u73B0/\u9A8C\u8BC1\u7ED3\u679C\uFF0C\u4E00\u952E\u91CD\u7B97\u5E76\u6301\u4E45\u5316\u95E8\u7981\u8BC1\u4E66") : t("\u5F53\u524D\u4F1A\u8BDD\u6CA1\u6709\u53EF\u91CD\u7B97\u7684\u95E8\u7981\u6570\u636E\uFF08\u9700\u5148\u6709\u5BA1\u67E5/\u9A8C\u8BC1\u7ED3\u679C\uFF09")
+      }, copiedKey === "qgate-compute" ? t("\u5DF2\u590D\u5236") : t("\u91CD\u7B97\u95E8\u7981")),
       React.createElement("button", {
         className: "iterate-btn",
         "data-danger": "",

package/lib/parse.js

@@ -1967,6 +1967,198 @@ export function buildQualityGateClearInstruction() {
   )
 }
 
+/**
+ * `iterate_quality_gate` compute payload — the one-click "recompute the gate
+ * from THIS session" builder.
+ *
+ * The gate's `compute` operation needs a hand-shaped payload
+ * (`dimensions` / `findings` / `validationResults` / `findingsByRound` /
+ * `fixedByDimension`) that a human cannot reliably type by hand — especially
+ * the per-dimension per-round convergence series. This builder turns the
+ * observatory's manifest (already normalized into the session stream) into
+ * that payload, so the model applies the exact data the user is looking at.
+ *
+ * Every field is optional and only included when non-empty (an empty findings
+ * list is still a legal `compute` — it yields a `pending` certificate, never
+ * a fabricated FAIL — so a caller may still emit a bare `{ operation }`).
+ * The returned text is a paste-able natural-language instruction, mirroring
+ * the rest of the panel's copy-to-command payloads.
+ *
+ * @param {{
+ *   dimensions?: string[],
+ *   findings?: Array<{ dimension: string, severity: string, file: string, line?: number }>,
+ *   validationResults?: Array<{ command: string, exitCode: number }>,
+ *   findingsByRound?: Record<string, number[]>,
+ *   fixedByDimension?: Record<string, number>,
+ * } | null | undefined} input
+ * @returns {string}
+ */
+export function buildQualityGateComputeInstruction(input) {
+  const src = input && typeof input === 'object' ? input : {}
+  const cleanStr = (v) => typeof v === 'string' && v.length > 0
+  const dimensions = Array.isArray(src.dimensions)
+    ? src.dimensions.filter((d) => typeof d === 'string' && d.length > 0)
+    : []
+  const findings = Array.isArray(src.findings)
+    ? src.findings.filter((f) => f && typeof f === 'object' && cleanStr(f.dimension) && cleanStr(f.file))
+        .map((f) => ({
+          dimension: f.dimension,
+          severity: typeof f.severity === 'string' && f.severity ? f.severity : 'medium',
+          file: f.file,
+          ...(typeof f.line === 'number' && Number.isFinite(f.line) && f.line > 0 ? { line: f.line } : {}),
+        }))
+    : []
+  const validationResults = Array.isArray(src.validationResults)
+    ? src.validationResults
+        .filter((v) => v && typeof v === 'object' && cleanStr(v.command) && typeof v.exitCode === 'number' && Number.isFinite(v.exitCode))
+        .map((v) => ({ command: v.command, exitCode: v.exitCode }))
+    : []
+  const findingsByRound = {}
+  if (src.findingsByRound && typeof src.findingsByRound === 'object' && !Array.isArray(src.findingsByRound)) {
+    for (const [dim, series] of Object.entries(src.findingsByRound)) {
+      const clean = Array.isArray(series)
+        ? series.filter((n) => typeof n === 'number' && Number.isFinite(n) && n >= 0)
+        : []
+      if (clean.length > 0) findingsByRound[dim] = clean
+    }
+  }
+  const fixedByDimension = {}
+  if (src.fixedByDimension && typeof src.fixedByDimension === 'object' && !Array.isArray(src.fixedByDimension)) {
+    for (const [dim, n] of Object.entries(src.fixedByDimension)) {
+      if (typeof n === 'number' && Number.isFinite(n) && n >= 0) fixedByDimension[dim] = n
+    }
+  }
+  const payload = { operation: 'compute' }
+  if (dimensions.length > 0) payload.dimensions = dimensions
+  if (findings.length > 0) payload.findings = findings
+  if (validationResults.length > 0) payload.validationResults = validationResults
+  if (Object.keys(findingsByRound).length > 0) payload.findingsByRound = findingsByRound
+  if (Object.keys(fixedByDimension).length > 0) payload.fixedByDimension = fixedByDimension
+  return (
+    '请调用 `iterate_quality_gate` 重算质量门禁证书（从当前会话推导维度/发现/验证结果并持久化）：\n\n' +
+    '```json\n' +
+    `${JSON.stringify(payload, null, 2)}\n` +
+    '```'
+  )
+}
+
+/**
+ * Reduce a normalized transcript manifest (+ the session-scanned quality gate)
+ * into the `iterate_quality_gate compute` input. Pure: reads only the fields
+ * the observatory already renders, so the one-click recompute can never
+ * disagree with what the user is looking at.
+ *
+ * @param {Record<string, unknown> | null | undefined} manifest normalized transcript
+ * @param {{ dimensions?: Array<{ dimension?: string }> } | null | undefined} qualityGate
+ * @returns {{
+ *   dimensions: string[],
+ *   findings: Array<{ dimension: string, severity: string, file: string, line?: number }>,
+ *   validationResults: Array<{ command: string, exitCode: number }>,
+ *   findingsByRound: Record<string, number[]>,
+ *   fixedByDimension: Record<string, number>,
+ * }}
+ */
+export function deriveQualityGateComputeInput(manifest, qualityGate) {
+  const m = manifest && typeof manifest === 'object' ? manifest : {}
+  const asArray = (v) => (Array.isArray(v) ? v : [])
+
+  // dimensions: union of every thread dimension across rounds, falling back
+  // to the already-computed gate's dimension names (a dry-run with no threads
+  // but a previous certificate still recomputes with the same gated set).
+  const dims = []
+  for (const r of asArray(m.rounds)) {
+    for (const t of asArray(r && r.threads)) {
+      const d = t && typeof t.dimension === 'string' ? t.dimension : ''
+      if (d && !dims.includes(d)) dims.push(d)
+    }
+  }
+  if (dims.length === 0 && qualityGate && Array.isArray(qualityGate.dimensions)) {
+    for (const d of qualityGate.dimensions) {
+      const name = d && typeof d.dimension === 'string' ? d.dimension : ''
+      if (name && !dims.includes(name)) dims.push(name)
+    }
+  }
+
+  const findings = []
+  for (const f of asArray(m.findings)) {
+    if (!f || typeof f !== 'object') continue
+    if (typeof f.dimension !== 'string' || !f.dimension) continue
+    if (typeof f.file !== 'string' || !f.file) continue
+    findings.push({
+      dimension: f.dimension,
+      severity: typeof f.severity === 'string' && f.severity ? f.severity : 'medium',
+      file: f.file,
+      ...(typeof f.line === 'number' && Number.isFinite(f.line) && f.line > 0 ? { line: f.line } : {}),
+    })
+  }
+
+  const validationResults = []
+  for (const v of asArray(m.validations)) {
+    if (!v || typeof v !== 'object') continue
+    if (typeof v.command !== 'string' || !v.command) continue
+    if (typeof v.exitCode !== 'number' || !Number.isFinite(v.exitCode)) continue
+    validationResults.push({ command: v.command, exitCode: v.exitCode })
+  }
+
+  // findingsByRound: per dimension, per review round, the number of findings
+  // raised that round — a true convergence series (latest last), padded to the
+  // round count so the gate's convergenceRate measures the actual first→last
+  // shrink even when a dimension is absent from a round's threads (0).
+  const rounds = asArray(m.rounds)
+  const allDims = dims.slice()
+  for (const r of rounds) {
+    for (const t of asArray(r && r.threads)) {
+      const d = t && typeof t.dimension === 'string' ? t.dimension : ''
+      if (d && !allDims.includes(d)) allDims.push(d)
+    }
+  }
+  const byRound = {}
+  for (const d of allDims) {
+    byRound[d] = []
+  }
+  for (const r of rounds) {
+    for (const d of allDims) {
+      let n = 0
+      for (const t of asArray(r && r.threads)) {
+        if (!t || t.dimension !== d) continue
+        for (const f of asArray(t && t.findings)) {
+          if (f && typeof f === 'object' && typeof f.dimension === 'string' && f.dimension === d) n += 1
+        }
+      }
+      byRound[d].push(n)
+    }
+  }
+
+  // fixedByDimension: attribute successful fixes to a dimension by matching
+  // the fixed file (and line, when both sides carry one) to a finding. A fix
+  // with no match is left unattributed — the field is optional and a gate
+  // recompute stays correct without it.
+  const fixedByDimension = {}
+  const findingsIndex = findings
+  for (const fx of asArray(m.fixes)) {
+    if (!fx || typeof fx !== 'object' || fx.success !== true) continue
+    const file = typeof fx.file === 'string' ? fx.file : ''
+    if (!file) continue
+    let dim = ''
+    for (const f of findingsIndex) {
+      if (f.file !== file) continue
+      if (typeof f.line === 'number' && typeof fx.line === 'number' && f.line !== fx.line) continue
+      dim = f.dimension
+      break
+    }
+    if (!dim) continue
+    fixedByDimension[dim] = (fixedByDimension[dim] || 0) + 1
+  }
+
+  return {
+    dimensions: dims,
+    findings,
+    validationResults,
+    findingsByRound: byRound,
+    fixedByDimension,
+  }
+}
+
 /**
  * `iterate_preflight` readiness-check payload (runs the pre-run checklist).
  *

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "iterate-plugin",
-  "version": "3.16.0",
+  "version": "3.17.0",
   "description": "dsh plugin that turns the iterate skill into an autonomous closed-loop harness with quality command center and experience bank (v3.3). Features: plan -> parallel review xN -> atomic fixes -> validate -> loop -> auto-stop, plus dry-run pure-review mode, quality gate compute/persist, writable experience bank, defense events stream (record + bilingual labels), and native command buttons — with a live quality command center (F8/F9/F10 render real session data), §8 assign-fix instruction, and task_mode indicator wired end to end.",
   "type": "module",
   "license": "MIT",

package/src/client/i18n.ts

@@ -317,6 +317,9 @@ export const EN: Record<string, string> = {
   基准: 'Baseline',
   当前: 'Current',
   变化: 'Delta',
+  复制对比表: 'Copy comparison table',
+  '复制跨会话对比表为 Markdown（可粘贴进 PR / issue / CI 注释）':
+    'Copy the cross-session comparison as a Markdown table (paste into a PR / issue / CI comment)',
 
   // ── F3 finding locations ─────────────────────────────────────────────────
   暂无发现: 'No findings',
@@ -398,6 +401,11 @@ export const EN: Record<string, string> = {
   '复制 iterate_quality_gate clear 指令文本（重置陈旧的 FAIL 证书）':
     'Copy the iterate_quality_gate clear instruction (resets a stale FAIL certificate)',
   清除门禁: 'Clear gate',
+  重算门禁: 'Recompute gate',
+  '复制 iterate_quality_gate compute 指令：从当前会话推导维度/发现/验证结果，一键重算并持久化门禁证书':
+    'Copy the iterate_quality_gate compute instruction: derive dimensions / findings / validation results from this session and recompute + persist the gate certificate',
+  '当前会话没有可重算的门禁数据（需先有审查/验证结果）':
+    'No gate data in this session to recompute (need review / validation results first)',
   '复制拉取磁盘快照指令：让模型回显 .iterate/ 下的门禁、经验、防御事件与 transcript':
     'Copy the disk-snapshot instruction: have the model echo the gate, experience, defense events and transcript artifacts under .iterate/',
   质量门禁视图: 'Quality gate view',

package/src/client/index.ts

@@ -101,6 +101,8 @@ import {
   buildRollbackInstruction, // #10: shared rollback payload
   buildQualityGateQueryInstruction, // #10: JSON-carrying query instruction
   buildQualityGateClearInstruction, // #10: JSON-carrying clear instruction
+  buildQualityGateComputeInstruction, // v3.17: one-click recompute payload from THIS session
+  deriveQualityGateComputeInput, // v3.17: manifest → compute input reducer
   buildTriageReadbackInstruction, // #7: close the write-back loop
   buildExperienceListInstruction, // #3/#10: JSON list payload for F9
   buildDefenseEventsListInstruction, // #3/#10: JSON list payload for F10
@@ -153,6 +155,7 @@ import {
   buildRunBaseline,
   parseRunBaseline,
   compareRuns,
+  buildRunCompareMarkdown,
   RUN_BASELINE_STORAGE_KEY,
   type RunBaseline,
   type RunCompareMetric,
@@ -2920,6 +2923,21 @@ function ObservatoryPanel(props: SlotProps) {
                 onClick: () => setShowCompare((v) => !v),
               }, showCompare ? t('收起对比') : t('与基准对比'))
             : null,
+          baseline && current && showCompare
+            ? React.createElement('button', {
+                className: 'iterate-btn', 'data-copied': copiedKey === 'rc-copy' ? '' : undefined,
+                onClick: () => copyInstruction('rc-copy', buildRunCompareMarkdown(compareRuns(baseline, current), {
+                  指标: t('指标'),
+                  基准: t('基准'),
+                  当前: t('当前'),
+                  变化: t('变化'),
+                  ...Object.fromEntries(
+                    Object.entries(RUN_COMPARE_METRIC_LABELS).map(([k, v]) => [k, v[getLang()]]),
+                  ),
+                })),
+                title: t('复制跨会话对比表为 Markdown（可粘贴进 PR / issue / CI 注释）'),
+              }, copiedKey === 'rc-copy' ? t('已复制') : t('复制对比表'))
+            : null,
           baseline
             ? React.createElement('button', {
                 className: 'iterate-btn', 'data-ghost': '',
@@ -3327,6 +3345,19 @@ function ObservatoryPanel(props: SlotProps) {
     // Reset a stale FAIL certificate before a fresh iteration — the tool's
     // `clear` operation existed since v3.5.1 but had no client entry point.
     const gateClearInstruction = buildQualityGateClearInstruction()
+    // v3.17: one-click recompute. The gate's `compute` payload needs a
+    // hand-shaped `findingsByRound` convergence series that a human cannot
+    // type by hand — this derives it from the session manifest the panel is
+    // already rendering, so the recompute can never disagree with the view.
+    const computeInput = deriveQualityGateComputeInput(
+      manifest as unknown as Record<string, unknown> | null | undefined,
+      qualityGate as unknown as Record<string, unknown> | null | undefined,
+    )
+    const computeInstruction = buildQualityGateComputeInstruction(computeInput)
+    const hasComputeData =
+      computeInput.dimensions.length > 0 ||
+      computeInput.findings.length > 0 ||
+      computeInput.validationResults.length > 0
     const gate = qualityGate
     const dims = gate && gate.dimensions ? gate.dimensions : []
     const hasGate = gate !== null && Boolean(gate.overallStatus || gate.overallScore != null || dims.length > 0)
@@ -3342,6 +3373,14 @@ function ObservatoryPanel(props: SlotProps) {
         onClick: () => copyInstruction('qgate', gateInstruction),
         title: t('复制 iterate_quality_gate 查询指令（含 JSON 参数）'),
       }, copiedKey === 'qgate' ? t('已复制') : t('查询门禁')),
+      React.createElement('button', {
+        className: 'iterate-btn', 'data-primary': '', 'data-copied': copiedKey === 'qgate-compute' ? '' : undefined,
+        disabled: !hasComputeData,
+        onClick: () => copyInstruction('qgate-compute', computeInstruction),
+        title: hasComputeData
+          ? t('复制 iterate_quality_gate compute 指令：从当前会话推导维度/发现/验证结果，一键重算并持久化门禁证书')
+          : t('当前会话没有可重算的门禁数据（需先有审查/验证结果）'),
+      }, copiedKey === 'qgate-compute' ? t('已复制') : t('重算门禁')),
       React.createElement('button', {
         className: 'iterate-btn', 'data-danger': '', 'data-copied': copiedKey === 'qgate-clear' ? '' : undefined,
         onClick: () => copyInstruction('qgate-clear', gateClearInstruction),

package/src/client/run-compare.ts

@@ -211,3 +211,29 @@ export function compareRuns(baseline: RunBaseline, current: RunBaseline): RunCom
   ]
   return rows
 }
+
+/**
+ * Render the side-by-side comparison as a Markdown table that can be pasted
+ * into a PR / issue / CI comment. `labels` is resolved by the caller from the
+ * current UI language (this module stays dependency-free and Chinese-free).
+ * A `null` cell renders as `—` (that side had no such measurement).
+ *
+ * @param rows - the rows from {@link compareRuns}
+ * @param labels - metric key → localized column name
+ */
+export function buildRunCompareMarkdown(rows: RunCompareMetric[], labels: Record<string, string>): string {
+  const cell = (v: number | null): string => (v === null ? '—' : String(v))
+  const header = ['指标', '基准', '当前', '变化'].map((h) => labels[h] ?? h)
+  const lines = [
+    `| ${header.join(' | ')} |`,
+    `| ${header.map(() => '---').join(' | ')} |`,
+  ]
+  for (const row of rows) {
+    const delta =
+      row.baseline === null || row.current === null ? '—' : String(row.current - row.baseline)
+    lines.push(
+      `| ${labels[row.key] ?? row.key} | ${cell(row.baseline)} | ${cell(row.current)} | ${delta} |`,
+    )
+  }
+  return lines.join('\n')
+}

package/src/index.ts

@@ -2,16 +2,17 @@
  * iterate-plugin — dsh plugin for the iterate autonomous closed-loop workflow
  *
  * Architecture:
- * - The plugin registers 18 tools (14 original + 3 v3.0 quality command center
- *   tools + 1 v3.9 report export)
+ * - The plugin registers 19 tools (14 original + 3 v3.0 quality command center
+ *   tools + 1 v3.9 report export + 1 v3.12 pre-run preflight)
  *   Original: config, validate, decision-log, context, review, triage, fix, diff,
  *   rollback, checkpoint, status, history, prune, transcript
  *   v3.0: experience, quality_gate, defense_events
  *   v3.9: report
+ *   v3.12: preflight
  * - The plugin injects a system prompt section teaching the iterate workflow pattern
  * - The model (prompted by the skill) writes a workflow script using dsh's `workflow` tool
  * - The workflow script uses `agent()` / `parallel()` / `phase()` / `log()` to orchestrate
- * - Subagents use the 18 tools to do real work (read config, run validation, log decisions,
+ * - Subagents use the 19 tools to do real work (read config, run validation, log decisions,
  *   review, triage, apply/rollback/fixing, checkpoint, status, history, prune, transcript,
  *   query experience bank, check quality gates, query defense events, export a report)
  * - A `tools/pre-execute` hook gates destructive iterate calls behind human approval
@@ -25,7 +26,7 @@
  *
  * Key files:
  * - src/index.ts      — Plugin entry: register tools + inject skill prompt
- * - src/tools/        — 18 tool implementations (14 original + 3 v3.0 + 1 v3.9)
+ * - src/tools/        — 19 tool implementations (14 original + 3 v3.0 + 1 v3.9 + 1 v3.12)
  * - src/config-loader.ts — YAML config loading
  * - src/types.ts     — Shared types
  */

package/src/tools/decision-log.ts

@@ -323,6 +323,25 @@ function decodedLines(filePath: string): Array<{ raw: string; decoded: unknown }
   return out
 }
 
+/**
+ * Rebuild the exact ledger text the kernel verifies, dropping only the artifact a
+ * trailing newline leaves. Unlike `decodedLines` this keeps interior blank lines:
+ * `kernelVerifyLedgerText` splits on "\n" and `JSON.parse`s each line, so an interior
+ * blank line is a corrupt entry it must report — renumbering the lines that follow it
+ * (by dropping it here) would mask a real break as a healthy ledger.
+ */
+function ledgerTextForVerify(filePath: string): string {
+  let text: string
+  try {
+    text = readFileSync(filePath, 'utf-8')
+  } catch {
+    return ''
+  }
+  const lines = text.endsWith('\n') ? text.slice(0, -1) : text
+  if (lines.length === 0) return ''
+  return `${lines}\n`
+}
+
 /** The 1-based line number of the first legacy record AFTER the chain began, else `null`. */
 export function contestedLedgerLine(filePath: string): number | null {
   const kinds = decodedLines(filePath).map((l) => classifyDecisionLine(l.decoded))
@@ -798,8 +817,7 @@ export function verifyDecisionLedger(projectRoot: string): {
 } {
   const filePath = decisionLogPath(projectRoot)
   try {
-    const lines = decodedLines(filePath)
-    const body = `${lines.map((l) => l.raw).join('\n')}${lines.length > 0 ? '\n' : ''}`
+    const body = ledgerTextForVerify(filePath)
     const verified = kernelVerifyLedgerText(body)
     return {
       ok: verified.ok,