superlab 0.1.40 → 0.1.42

package/lib/i18n.cjs

@@ -473,6 +473,7 @@ const ZH_SKILL_FILES = {
 - 不要把 idea 工件本身当成唯一证据记录；两轮文献检索的查询、来源分桶和最终来源数必须同步记到 \`.lab/writing/idea-source-log.md\`。
 - 在 \`idea\` 阶段可以说大致怎么验证、最小实验是什么，但不要在这里冻结 sample size、招募方案、条件数、问卷设计或随机化 protocol。
 - human-subject experiment design 应该留到 \`/lab:spec\`，在那里再把招募、分组、测量和伦理细节写死。
+- 使用 \`skills/lab/references/recipes.md\` 作为常见阶段链路的快速路径，不要据此发明新的命令或别名。
 - 三个 meaningful points 每个都控制在一句直接的话里。
 - 在批准前，必须运行 \`.lab/.managed/scripts/validate_idea_artifact.py --idea <idea-artifact> --source-log .lab/writing/idea-source-log.md --workflow-config .lab/config/workflow.json\`。
 - rewrite-only 模式下不能更新 \`.lab/context/mission.md\`、\`.lab/context/decisions.md\` 或 \`.lab/context/open-questions.md\`。
@@ -518,6 +519,46 @@ const ZH_SKILL_FILES = {
 - 主指标是否符合顶会惯例
 - claims 是否有记录在案的证据支撑
 - failures 和 limitations 是否被保留
+`,
+  [path.join(".codex", "skills", "lab", "references", "recipes.md")]:
+`# /lab 阶段快速路径
+
+把这个文件当成常见阶段链路的快速路由图。它不会新增命令，也不会替代各 stage 的正式合同。
+
+## idea -> data -> spec
+
+当问题还在筛选或收窄时，走这条路径。
+
+- 从 \`/lab:idea\` 开始，对候选方向做比较、文献检索，并在 approval gate 处收口。
+- idea 批准后，如果需要冻结 benchmark 方案，就进入 \`/lab:data\`。
+- idea 和 data package 都批准后，如果需要把变更写成可执行设计，就进入 \`/lab:spec\`。
+
+## run -> iterate -> review -> report
+
+当 mission 已批准、当前工作变成证据生成时，走这条路径。
+
+- 从 \`/lab:run\` 开始，先做最小有意义验证。
+- 当 campaign 需要在冻结 mission 内做 bounded rounds 时，进入 \`/lab:iterate\`。
+- 当方法学、公平性、泄漏风险或 claim discipline 需要 reviewer 式审查时，进入 \`/lab:review\`。
+- 当验证过的证据已经稳定到可以给协作者阅读时，进入 \`/lab:report\`。
+
+## framing -> write -> review
+
+当 report 工件已经稳定、论文叙事需要定稿时，走这条路径。
+
+- 从 \`/lab:framing\` 开始，锁定 paper-facing 名称、题目方向和 contribution wording。
+- 进入 \`/lab:write\`，基于已批准的 framing 和 report 证据逐 section 起草稿件。
+- 当 section 草稿或整篇论文的 claim set 需要 reviewer mode critique 时，回到 \`/lab:review\`。
+
+## \`/lab:auto\` 按已批准范围使用
+
+只有在上游阶段决策都已经批准后，才使用 \`/lab:auto\`。
+
+- \`L1\`：针对已批准执行阶段的一次 bounded validation cycle
+- \`L2\`：在冻结核心边界内做默认的 bounded experiment iteration
+- \`L3\`：更大范围的 campaign；只有 framing 已批准时才可选带写作
+
+\`/lab:auto\` 是对已批准阶段链路的编排层，不替代 \`idea\`、\`data\`、\`framing\` 或 \`spec\`。
 `,
   [path.join(".lab", ".managed", "templates", "idea.md")]:
 `# Idea 工件
@@ -1541,6 +1582,11 @@ const ZH_SKILL_FILES = {
 - Terminal goal type:
 - Terminal goal target:
 - Required terminal artifact:
+- Primary gate:
+- Secondary guard:
+- Promotion condition:
+- Stop reason:
+- Escalation reason:
 - 如果 workflow language 是中文，摘要、清单条目、任务标签和进度更新都应使用中文。
 - 示例 Objective: 推进 paper layer 3，完成一轮 bounded protocol、测试、最小实现和一轮小规模结果。
 
@@ -2083,7 +2129,7 @@ ZH_CONTENT[path.join(".lab", ".managed", "templates", "framing.md")] = `# 论文
 ZH_CONTENT[path.join(".codex", "prompts", "lab.md")] = codexPrompt(
   "查看 /lab 研究工作流总览并选择合适阶段",
   "workflow question 或 stage choice",
-  "# `/lab` for Codex\n\n`/lab` 是严格的研究工作流命令族。每次都使用同一套仓库工件和阶段边界。\n\n## 子命令\n\n- `/lab:idea`\n  先做两轮脑暴和两轮文献检索，再定义问题与 failure case、对比最接近前作，并输出带 approval gate 的 source-backed recommendation。\n\n- `/lab:data`\n  把已批准的 idea 转成数据集与 benchmark 方案，记录数据集年份、使用过该数据集的论文、下载来源、许可或访问限制，以及 classic-public、recent-strong-public、claim-specific 三类 benchmark 的纳入理由，和 canonical baselines、strong historical baselines、recent strong public methods、closest prior work 四类对比方法的纳入理由。\n\n- `/lab:auto`\n  在不改变 mission、framing 和核心 claims 的前提下，读取 eval-protocol 与 auto-mode 契约并自动编排 `run`、`iterate`、`review`、`report`，必要时扩展数据集、benchmark 和 comparison methods，并在满足升格策略时自动升级 primary package。启动前必须选定 autonomy level、声明 terminal goal，并显式批准契约。\n\n- `/lab:framing`\n  通过审计当前领域与相邻领域的术语，锁定 paper-facing 的方法名、模块名、论文题目和 contribution bullets，并在 section 起草前保留 approval gate。\n\n- `/lab:spec`\n  把已批准的 idea 转成 `.lab/changes/<change-id>/` 下的一个 lab change 目录，并在其中写出 `proposal`、`design`、`spec`、`tasks`。\n\n- `/lab:run`\n  执行最小有意义验证运行，登记 run，并生成第一版标准化评估摘要。\n\n- `/lab:iterate`\n  在冻结 mission、阈值、verification commands 与 `completion_promise` 的前提下执行有边界的实验迭代。\n\n- `/lab:review`\n  以 reviewer mode 审查文档或结果，先给短摘要，再输出 findings、fatal flaws、fix priority 和 residual risks。\n\n- `/lab:report`\n  从 runs 和 iterations 工件生成最终研究报告。\n\n- `/lab:write`\n  使用已安装 `lab` skill 下 vendored 的 paper-writing references，把稳定 report 工件转成论文 section。\n\n## 调度规则\n\n- 始终使用 `skills/lab/SKILL.md` 作为工作流合同。\n- 用户显式调用 `/lab:<stage>` 时，要立刻执行该 stage，而不是只推荐别的 `/lab` stage。\n- 先给简洁的阶段摘要；只要 stage 合同要求受管工件，就应立刻落盘，再回报输出路径和下一步。\n- 如果歧义会影响结论，一次只问一个问题；如果有多条可行路径，先给 2-3 个方案再收敛。\n- `/lab:spec` 前应已有经批准的数据集与 benchmark 方案。\n- `/lab:run`、`/lab:iterate`、`/lab:auto`、`/lab:report` 都应遵循 `.lab/context/eval-protocol.md`。\n- `.lab/context/eval-protocol.md` 不只定义主指标和主表，也应定义指标释义、实验阶梯，以及指标和对比实现的来源。\n- `/lab:auto` 只编排已批准边界内的执行阶段，不替代手动的 idea/data/framing/spec 决策。\n- `/lab:write` 前必须已有经批准的 `/lab:framing` 工件。\n\n## 如何输入 `/lab:auto`\n\n## `/lab:auto` 层级指南\n\n- `L1`：适合安全验证、一轮 bounded 真实运行，或简单 report 刷新。\n- `L2`：默认推荐级别，适合冻结核心边界内的常规实验迭代。\n- `L3`：激进 campaign 级别，只在你明确想做更大范围探索和可选写作时使用。\n- 如果不确定，默认推荐 `L2`。\n- 如果用户输入没写级别，或者把级别和 `paper layer`、`phase`、`table` 混用了，就应先停下来，要求用户明确选 `L1/L2/L3`。\n\n- 把 `Autonomy level L1/L2/L3` 视为执行权限级别，不要和论文里的 layer、phase、table 编号混用。\n- 把 `paper layer`、`phase`、`table` 视为实验目标。例如 `paper layer 3` 或 `Phase 1` 不是 `Autonomy level L3`。\n- 一条好的 `/lab:auto` 输入应至少说清：objective、自治级别、terminal goal、scope、allowed modifications。\n- 如果 workflow language 是中文，摘要、清单条目、任务标签和进度更新都应使用中文，除非文件路径、代码标识符或字面指标名必须保持原样。\n- 示例：`/lab:auto 自治级别 L2。目标：推进 paper layer 3。终止条件：完成 bounded protocol、测试、最小实现和一轮小规模结果。允许修改：配置、数据接入、评估脚本。`\n"
+  "# `/lab` for Codex\n\n`/lab` 是严格的研究工作流命令族。每次都使用同一套仓库工件和阶段边界。\n\n## 子命令\n\n- `/lab:idea`\n  先做两轮脑暴和两轮文献检索，再定义问题与 failure case、对比最接近前作，并输出带 approval gate 的 source-backed recommendation。\n\n- `/lab:data`\n  把已批准的 idea 转成数据集与 benchmark 方案，记录数据集年份、使用过该数据集的论文、下载来源、许可或访问限制，以及 classic-public、recent-strong-public、claim-specific 三类 benchmark 的纳入理由，和 canonical baselines、strong historical baselines、recent strong public methods、closest prior work 四类对比方法的纳入理由。\n\n- `/lab:auto`\n  在不改变 mission、framing 和核心 claims 的前提下，读取 eval-protocol 与 auto-mode 契约并自动编排 `run`、`iterate`、`review`、`report`，必要时扩展数据集、benchmark 和 comparison methods，并在满足升格策略时自动升级 primary package。启动前必须选定 autonomy level、声明 terminal goal，并显式写清 primary gate、secondary guard、promotion condition、stop reason 和 escalation reason，再批准契约。\n\n- `/lab:framing`\n  通过审计当前领域与相邻领域的术语，锁定 paper-facing 的方法名、模块名、论文题目和 contribution bullets，并在 section 起草前保留 approval gate。\n\n- `/lab:spec`\n  把已批准的 idea 转成 `.lab/changes/<change-id>/` 下的一个 lab change 目录，并在其中写出 `proposal`、`design`、`spec`、`tasks`。\n\n- `/lab:run`\n  执行最小有意义验证运行，登记 run，并生成第一版标准化评估摘要。\n\n- `/lab:iterate`\n  在冻结 mission、阈值、verification commands 与 `completion_promise` 的前提下执行有边界的实验迭代。\n\n- `/lab:review`\n  以 reviewer mode 审查文档或结果，先给短摘要，再输出 findings、fatal flaws、fix priority 和 residual risks。\n\n- `/lab:report`\n  从 runs 和 iterations 工件生成最终研究报告。\n\n- `/lab:write`\n  使用已安装 `lab` skill 下 vendored 的 paper-writing references，把稳定 report 工件转成论文 section。\n\n## 调度规则\n\n- 始终使用 `skills/lab/SKILL.md` 作为工作流合同。\n- 用户显式调用 `/lab:<stage>` 时，要立刻执行该 stage，而不是只推荐别的 `/lab` stage。\n- 先给简洁的阶段摘要；只要 stage 合同要求受管工件，就应立刻落盘，再回报输出路径和下一步。\n- 如果歧义会影响结论，一次只问一个问题；如果有多条可行路径，先给 2-3 个方案再收敛。\n- `/lab:spec` 前应已有经批准的数据集与 benchmark 方案。\n- `/lab:run`、`/lab:iterate`、`/lab:auto`、`/lab:report` 都应遵循 `.lab/context/eval-protocol.md`。\n- `.lab/context/eval-protocol.md` 不只定义主指标和主表，也应定义指标释义、实验阶梯，以及指标和对比实现的来源。\n- `/lab:auto` 只编排已批准边界内的执行阶段，不替代手动的 idea/data/framing/spec 决策。\n- `/lab:write` 前必须已有经批准的 `/lab:framing` 工件。\n\n## 如何输入 `/lab:auto`\n\n## `/lab:auto` 层级指南\n\n- `L1`：适合安全验证、一轮 bounded 真实运行，或简单 report 刷新。\n- `L2`：默认推荐级别，适合冻结核心边界内的常规实验迭代。\n- `L3`：激进 campaign 级别，只在你明确想做更大范围探索和可选写作时使用。\n- 如果不确定，默认推荐 `L2`。\n- 如果用户输入没写级别，或者把级别和 `paper layer`、`phase`、`table` 混用了，就应先停下来，要求用户明确选 `L1/L2/L3`。\n- 真正的 `/lab:auto` 首个可见输出必须是 `Auto preflight`。\n- 这个首个可见输出必须展示已读取文件，以及 `Autonomy level`、`Allowed stages`、`Terminal goal`、`Primary gate`、`Secondary guard`。\n- 如果无法从 `.lab/context/eval-protocol.md`、`.lab/context/auto-mode.md`、`.lab/context/auto-status.md` 和 `.lab/context/auto-outcome.md` 完成 preflight，就必须停下，而不是假装 loop 已经 armed。\n\n- 把 `Autonomy level L1/L2/L3` 视为执行权限级别，不要和论文里的 layer、phase、table 编号混用。\n- 把 `paper layer`、`phase`、`table` 视为实验目标。例如 `paper layer 3` 或 `Phase 1` 不是 `Autonomy level L3`。\n- 一条好的 `/lab:auto` 输入应至少说清：objective、自治级别、terminal goal、scope、allowed modifications。\n- 如果 workflow language 是中文，摘要、清单条目、任务标签和进度更新都应使用中文，除非文件路径、代码标识符或字面指标名必须保持原样。\n- 示例：`/lab:auto 自治级别 L2。目标：推进 paper layer 3。终止条件：完成 bounded protocol、测试、最小实现和一轮小规模结果。允许修改：配置、数据接入、评估脚本。`\n"
 );
 
 ZH_CONTENT[path.join(".codex", "prompts", "lab-data.md")] = codexPrompt(
@@ -2095,14 +2141,14 @@ ZH_CONTENT[path.join(".codex", "prompts", "lab-data.md")] = codexPrompt(
 ZH_CONTENT[path.join(".codex", "prompts", "lab-auto.md")] = codexPrompt(
   "在已批准边界内编排自动实验循环",
   "auto mode objective",
-  "使用已安装的 `lab` 技能：`.codex/skills/lab/SKILL.md`。\n\n立刻针对用户当前给出的参数执行 `/lab:auto`，不要只推荐别的 `/lab` 阶段。只有在缺少阻塞性前提时，才明确指出缺什么，并且一次最多追问一个问题。\n\n本命令运行 `/lab:auto` 阶段。它必须读取 `.lab/context/eval-protocol.md`、`.lab/context/auto-mode.md`、`.lab/context/auto-status.md` 与 `.lab/context/auto-outcome.md`，先确认 autonomy level、approval status 与 terminal goal schema，再把 eval-protocol 里的指标释义、主表计划、来源约束与结构化实验阶梯当作执行依据，在不修改 mission、framing 和核心 claims 的前提下编排已批准的 `run`、`iterate`、`review`、`report`，轮询长任务完成情况；如果声明了 rung，就保持会话活着并按 rung 转移继续推进。\n如果仓库的 workflow language 是中文，摘要、清单条目、任务标签和进度更新都必须使用中文，除非某个文件路径、代码标识符或字面指标名必须保持原样。\n把 `Layer 3`、`Phase 1`、`Table 2` 这类表达视为论文范围目标；只有显式写成 `Autonomy level L3` 或 `自治级别 L3` 时，才把它当成执行权限级别。\n不要用 `sleep 30`、单次 `pgrep` 或一次性的 `metrics.json` 探针来代替真实长任务命令；当真实实验进程还活着时，只允许发进度更新并继续等待。"
+  "使用已安装的 `lab` 技能：`.codex/skills/lab/SKILL.md`。\n\n立刻针对用户当前给出的参数执行 `/lab:auto`，不要只推荐别的 `/lab` 阶段。只有在缺少阻塞性前提时，才明确指出缺什么，并且一次最多追问一个问题。\n\n本命令运行 `/lab:auto` 阶段。它必须读取 `.lab/context/eval-protocol.md`、`.lab/context/auto-mode.md`、`.lab/context/auto-status.md` 与 `.lab/context/auto-outcome.md`，先确认 autonomy level、approval status、terminal goal schema，以及 primary gate、secondary guard、promotion condition、stop reason、escalation reason，再把 eval-protocol 里的指标释义、主表计划、来源约束与结构化实验阶梯当作执行依据，在不修改 mission、framing 和核心 claims 的前提下编排已批准的 `run`、`iterate`、`review`、`report`，轮询长任务完成情况；如果声明了 rung，就保持会话活着并按 rung 转移继续推进。\n首个可见输出块必须是 `Auto preflight`。这个块必须列出已读取文件，并回显 `Autonomy level`、`Approval status`、`Allowed stages`、`Terminal goal`、`Primary gate` 和 `Secondary guard`，然后才能进入执行摘要或动作计划。\n如果 preflight 所需字段缺失、过期或彼此冲突，就必须在执行前停下，并明确指出到底是哪一个字段阻止了 loop 启动。\n如果仓库的 workflow language 是中文，摘要、清单条目、任务标签和进度更新都必须使用中文，除非某个文件路径、代码标识符或字面指标名必须保持原样。\n把 `Layer 3`、`Phase 1`、`Table 2` 这类表达视为论文范围目标；只有显式写成 `Autonomy level L3` 或 `自治级别 L3` 时，才把它当成执行权限级别。\n不要用 `sleep 30`、单次 `pgrep` 或一次性的 `metrics.json` 探针来代替真实长任务命令；当真实实验进程还活着时，只允许发进度更新并继续等待。"
 );
 
 ZH_CONTENT[path.join(".claude", "commands", "lab.md")] = claudeCommand(
   "lab",
   "查看 /lab 研究工作流总览并选择合适阶段",
   "[stage] [target]",
-  "# `/lab` for Claude\n\n`/lab` 是 Claude Code 里的 lab 工作流分发入口。调用方式有两种：\n\n- `/lab <stage> ...`\n- `/lab-idea`、`/lab-data`、`/lab-auto`、`/lab-framing`、`/lab-spec`、`/lab-run`、`/lab-iterate`、`/lab-review`、`/lab-report`、`/lab-write`\n\n## 阶段别名\n\n- `/lab idea ...` 或 `/lab-idea`\n- `/lab data ...` 或 `/lab-data`\n- `/lab auto ...` 或 `/lab-auto`\n- `/lab framing ...` 或 `/lab-framing`\n- `/lab spec ...` 或 `/lab-spec`\n- `/lab run ...` 或 `/lab-run`\n- `/lab iterate ...` 或 `/lab-iterate`\n- `/lab review ...` 或 `/lab-review`\n- `/lab report ...` 或 `/lab-report`\n- `/lab write ...` 或 `/lab-write`\n\n## 调度规则\n\n- 始终使用 `skills/lab/SKILL.md` 作为工作流合同。\n- 用户显式调用 `/lab <stage> ...` 或 `/lab-<stage>` 时，要立刻执行该 stage，而不是只推荐别的阶段。\n- 先给简洁的阶段摘要；只要 stage 合同要求受管工件，就应立刻落盘，再回报输出路径和下一步。\n- 如果歧义会影响结论，一次只问一个问题；如果有多条可行路径，先给 2-3 个方案再收敛。\n- `spec` 前应已有经批准的数据集与 benchmark 方案。\n- `run`、`iterate`、`auto`、`report` 都应遵循 `.lab/context/eval-protocol.md`。\n- `auto` 只编排已批准边界内的执行阶段，不替代手动的 idea/data/framing/spec 决策。\n- `write` 前必须已有经批准的 `framing` 工件。\n\n## 如何输入 `/lab auto`\n\n## `/lab auto` 层级指南\n\n- `L1`：适合安全验证、一轮 bounded 真实运行，或简单 report 刷新。\n- `L2`：默认推荐级别，适合冻结核心边界内的常规实验迭代。\n- `L3`：激进 campaign 级别，只在你明确想做更大范围探索和可选写作时使用。\n- 如果不确定，默认推荐 `L2`。\n- 如果用户输入没写级别，或者把级别和 `paper layer`、`phase`、`table` 混用了，就应先停下来，要求用户明确选 `L1/L2/L3`。\n\n- 把 `Autonomy level L1/L2/L3` 视为执行权限级别，不要和论文里的 layer、phase、table 编号混用。\n- 把 `paper layer`、`phase`、`table` 视为实验目标。例如 `paper layer 3` 或 `Phase 1` 不是 `Autonomy level L3`。\n- 一条好的 `/lab auto` 输入应至少说清：objective、自治级别、terminal goal、scope、allowed modifications。\n- 如果 workflow language 是中文，摘要、清单条目、任务标签和进度更新都应使用中文，除非文件路径、代码标识符或字面指标名必须保持原样。\n- 示例：`/lab auto 自治级别 L2。目标：推进 paper layer 3。终止条件：完成 bounded protocol、测试、最小实现和一轮小规模结果。允许修改：配置、数据接入、评估脚本。`\n"
+  "# `/lab` for Claude\n\n`/lab` 是 Claude Code 里的 lab 工作流分发入口。调用方式有两种：\n\n- `/lab <stage> ...`\n- `/lab-idea`、`/lab-data`、`/lab-auto`、`/lab-framing`、`/lab-spec`、`/lab-run`、`/lab-iterate`、`/lab-review`、`/lab-report`、`/lab-write`\n\n## 阶段别名\n\n- `/lab idea ...` 或 `/lab-idea`\n- `/lab data ...` 或 `/lab-data`\n- `/lab auto ...` 或 `/lab-auto`\n- `/lab framing ...` 或 `/lab-framing`\n- `/lab spec ...` 或 `/lab-spec`\n- `/lab run ...` 或 `/lab-run`\n- `/lab iterate ...` 或 `/lab-iterate`\n- `/lab review ...` 或 `/lab-review`\n- `/lab report ...` 或 `/lab-report`\n- `/lab write ...` 或 `/lab-write`\n\n## 调度规则\n\n- 始终使用 `skills/lab/SKILL.md` 作为工作流合同。\n- 用户显式调用 `/lab <stage> ...` 或 `/lab-<stage>` 时，要立刻执行该 stage，而不是只推荐别的阶段。\n- 先给简洁的阶段摘要；只要 stage 合同要求受管工件，就应立刻落盘，再回报输出路径和下一步。\n- 如果歧义会影响结论，一次只问一个问题；如果有多条可行路径，先给 2-3 个方案再收敛。\n- `spec` 前应已有经批准的数据集与 benchmark 方案。\n- `run`、`iterate`、`auto`、`report` 都应遵循 `.lab/context/eval-protocol.md`。\n- `auto` 只编排已批准边界内的执行阶段，不替代手动的 idea/data/framing/spec 决策。\n- `write` 前必须已有经批准的 `framing` 工件。\n\n## 如何输入 `/lab auto`\n\n## `/lab auto` 层级指南\n\n- `L1`：适合安全验证、一轮 bounded 真实运行，或简单 report 刷新。\n- `L2`：默认推荐级别，适合冻结核心边界内的常规实验迭代。\n- `L3`：激进 campaign 级别，只在你明确想做更大范围探索和可选写作时使用。\n- 如果不确定，默认推荐 `L2`。\n- 如果用户输入没写级别，或者把级别和 `paper layer`、`phase`、`table` 混用了，就应先停下来，要求用户明确选 `L1/L2/L3`。\n- 真正的 `/lab auto` 首个可见输出必须是 `Auto preflight`。\n- 这个首个可见输出必须展示已读取文件，以及 `Autonomy level`、`Allowed stages`、`Terminal goal`、`Primary gate`、`Secondary guard`。\n- 如果无法从 `.lab/context/eval-protocol.md`、`.lab/context/auto-mode.md`、`.lab/context/auto-status.md` 和 `.lab/context/auto-outcome.md` 完成 preflight，就必须停下，而不是假装 loop 已经 armed。\n\n- 把 `Autonomy level L1/L2/L3` 视为执行权限级别，不要和论文里的 layer、phase、table 编号混用。\n- 把 `paper layer`、`phase`、`table` 视为实验目标。例如 `paper layer 3` 或 `Phase 1` 不是 `Autonomy level L3`。\n- 一条好的 `/lab auto` 输入应至少说清：objective、自治级别、terminal goal、scope、allowed modifications。\n- 如果 workflow language 是中文，摘要、清单条目、任务标签和进度更新都应使用中文，除非文件路径、代码标识符或字面指标名必须保持原样。\n- 示例：`/lab auto 自治级别 L2。目标：推进 paper layer 3。终止条件：完成 bounded protocol、测试、最小实现和一轮小规模结果。允许修改：配置、数据接入、评估脚本。`\n"
 );
 
 ZH_CONTENT[path.join(".claude", "commands", "lab-data.md")] = claudeCommand(
@@ -2116,7 +2162,24 @@ ZH_CONTENT[path.join(".claude", "commands", "lab-auto.md")] = claudeCommand(
   "lab-auto",
   "在已批准边界内编排自动实验循环",
   "auto mode objective",
-  "使用已安装的 `lab` 技能：`.claude/skills/lab/SKILL.md`。\n\n立刻针对用户当前给出的参数执行 `auto` 阶段，不要只推荐别的 lab 阶段。只有在缺少阻塞性前提时，才明确指出缺什么，并且一次最多追问一个问题。\n\n本命令运行 lab workflow 的 `auto` 阶段。它必须读取 `.lab/context/eval-protocol.md`、`.lab/context/auto-mode.md`、`.lab/context/auto-status.md` 与 `.lab/context/auto-outcome.md`，先确认 autonomy level、approval status 与 terminal goal schema，再把 eval-protocol 里的指标释义、主表计划、来源约束与结构化实验阶梯当作执行依据，在不修改 mission、framing 和核心 claims 的前提下编排已批准的 `run`、`iterate`、`review`、`report`，轮询长任务完成情况；如果声明了 rung，就保持会话活着并按 rung 转移继续推进。\n如果仓库的 workflow language 是中文，摘要、清单条目、任务标签和进度更新都必须使用中文，除非某个文件路径、代码标识符或字面指标名必须保持原样。\n把 `Layer 3`、`Phase 1`、`Table 2` 这类表达视为论文范围目标；只有显式写成 `Autonomy level L3` 或 `自治级别 L3` 时，才把它当成执行权限级别。\n不要用 `sleep 30`、单次 `pgrep` 或一次性的 `metrics.json` 探针来代替真实长任务命令；当真实实验进程还活着时，只允许发进度更新并继续等待。"
+  "使用已安装的 `lab` 技能：`.claude/skills/lab/SKILL.md`。\n\n立刻针对用户当前给出的参数执行 `auto` 阶段，不要只推荐别的 lab 阶段。只有在缺少阻塞性前提时，才明确指出缺什么，并且一次最多追问一个问题。\n\n本命令运行 lab workflow 的 `auto` 阶段。它必须读取 `.lab/context/eval-protocol.md`、`.lab/context/auto-mode.md`、`.lab/context/auto-status.md` 与 `.lab/context/auto-outcome.md`，先确认 autonomy level、approval status、terminal goal schema，以及 primary gate、secondary guard、promotion condition、stop reason、escalation reason，再把 eval-protocol 里的指标释义、主表计划、来源约束与结构化实验阶梯当作执行依据，在不修改 mission、framing 和核心 claims 的前提下编排已批准的 `run`、`iterate`、`review`、`report`，轮询长任务完成情况；如果声明了 rung，就保持会话活着并按 rung 转移继续推进。\n首个可见输出块必须是 `Auto preflight`。这个块必须列出已读取文件，并回显 `Autonomy level`、`Approval status`、`Allowed stages`、`Terminal goal`、`Primary gate` 和 `Secondary guard`，然后才能进入执行摘要或动作计划。\n如果 preflight 所需字段缺失、过期或彼此冲突，就必须在执行前停下，并明确指出到底是哪一个字段阻止了 loop 启动。\n如果仓库的 workflow language 是中文，摘要、清单条目、任务标签和进度更新都必须使用中文，除非某个文件路径、代码标识符或字面指标名必须保持原样。\n把 `Layer 3`、`Phase 1`、`Table 2` 这类表达视为论文范围目标；只有显式写成 `Autonomy level L3` 或 `自治级别 L3` 时，才把它当成执行权限级别。\n不要用 `sleep 30`、单次 `pgrep` 或一次性的 `metrics.json` 探针来代替真实长任务命令；当真实实验进程还活着时，只允许发进度更新并继续等待。"
+);
+
+const zhRecipeQuickPathLine =
+  "- 使用 `skills/lab/references/recipes.md` 作为常见阶段链路的快速路径，不要据此发明新的 slash 命令或别名。\n";
+
+ZH_CONTENT[path.join(".codex", "prompts", "lab.md")] = ZH_CONTENT[
+  path.join(".codex", "prompts", "lab.md")
+].replace(
+  "- 始终使用 `skills/lab/SKILL.md` 作为工作流合同。\n",
+  `- 始终使用 \`skills/lab/SKILL.md\` 作为工作流合同。\n${zhRecipeQuickPathLine}`
+);
+
+ZH_CONTENT[path.join(".claude", "commands", "lab.md")] = ZH_CONTENT[
+  path.join(".claude", "commands", "lab.md")
+].replace(
+  "- 始终使用 `skills/lab/SKILL.md` 作为工作流合同。\n",
+  `- 始终使用 \`skills/lab/SKILL.md\` 作为工作流合同。\n${zhRecipeQuickPathLine}`
 );
 
 ZH_CONTENT[path.join(".codex", "skills", "lab", "SKILL.md")] = `---
@@ -2637,8 +2700,15 @@ ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "auto.md")] = `# \`/la
 - 把 \`/lab:auto\` 当作编排层，不要再发明第二套 workflow。
 - 把 \`.lab/context/eval-protocol.md\` 当作论文导向指标、指标释义、主表、gate 与结构化实验阶梯的唯一来源。
 - 把评估协议当作“带来源的协议”，不是“临场想出来的说明”：指标定义、baseline 行为、对比实现和偏差都必须先写明来源，再用于 gate 或 promotion。
+- 把 \`.lab/context/auto-mode.md\` 当成可见的 control plane，在启动前显式写清 \`Primary gate\`、\`Secondary guard\`、\`Promotion condition\`、\`Stop reason\` 和 \`Escalation reason\`。
 - 契约里必须声明 \`Autonomy level\` 和 \`Approval status\`，只有显式写成 \`approved\` 才能启动。
 - 契约里还必须声明具体的 terminal goal：\`rounds\`、\`metric-threshold\` 或 \`task-completion\`，并补齐 \`Terminal goal target\` 与 \`Required terminal artifact\`。
+- control plane 术语固定为：
+  - \`Primary gate\`：当前 rung 或 loop 目标的主通过条件
+  - \`Secondary guard\`：即使主 gate 命中也能拦下通过的安全或异常检查
+  - \`Promotion condition\`：允许 exploratory 工作升格为 primary package 的明确条件
+  - \`Stop reason\`：结束当前循环但不升格的明确边界
+  - \`Escalation reason\`：必须转人工审查或收窄下一步的明确条件
 - 级别含义固定为：
   - \`L1\`：safe run，只允许 \`run\`、\`review\`、\`report\`
   - \`L2\`：bounded iteration，允许 \`run\`、\`iterate\`、\`review\`、\`report\`
@@ -2669,7 +2739,8 @@ ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "auto.md")] = `# \`/la
 5. 发起有边界动作
 6. 轮询进程、checkpoint 或 summary 的变化
 7. 评估声明过的 terminal goal 是否已经达成
-8. 记录结果并决定 continue、promote、stop 或 escalate
+8. 在正确边界上评估 \`Primary gate\`、\`Secondary guard\`、\`Promotion condition\`、\`Stop reason\` 和 \`Escalation reason\`
+9. 记录结果并决定 continue、promote、stop 或 escalate
 
 ## 交互约束
 

package/lib/install.cjs

@@ -520,6 +520,7 @@ function localizeInstalledAssets(targetDir, lang, { newlyCreatedProjectOwnedPath
     path.join(".codex", "skills", "lab", "stages", "report.md"),
     path.join(".codex", "skills", "lab", "stages", "write.md"),
     path.join(".codex", "skills", "lab", "references", "workflow.md"),
+    path.join(".codex", "skills", "lab", "references", "recipes.md"),
     path.join(".claude", "skills", "lab", "SKILL.md"),
     path.join(".claude", "skills", "lab", "stages", "idea.md"),
     path.join(".claude", "skills", "lab", "stages", "data.md"),
@@ -532,6 +533,7 @@ function localizeInstalledAssets(targetDir, lang, { newlyCreatedProjectOwnedPath
     path.join(".claude", "skills", "lab", "stages", "report.md"),
     path.join(".claude", "skills", "lab", "stages", "write.md"),
     path.join(".claude", "skills", "lab", "references", "workflow.md"),
+    path.join(".claude", "skills", "lab", "references", "recipes.md"),
     path.join(".lab", ".managed", "templates", "idea.md"),
     path.join(".lab", ".managed", "templates", "data.md"),
     path.join(".lab", ".managed", "templates", "framing.md"),

package/package-assets/claude/commands/lab-auto.md

@@ -7,7 +7,9 @@ argument-hint: autonomous campaign target
 Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
 
 Execute the requested `/lab-auto` command against the user's argument now. Do not only recommend another lab stage. If a blocking prerequisite is missing, say exactly what is missing and ask at most one clarifying question.
-This command runs the `auto` stage of the lab workflow. It must read `.lab/context/eval-protocol.md`, `.lab/context/auto-mode.md`, `.lab/context/auto-status.md`, and `.lab/context/auto-outcome.md`, enforce the declared terminal goal schema, orchestrate approved run, iterate, review, and report stages inside that contract, poll long-running work until completion or stop conditions, and write progress plus the final outcome back into `.lab/context/auto-status.md` and `.lab/context/auto-outcome.md`.
+This command runs the `auto` stage of the lab workflow. It must read `.lab/context/eval-protocol.md`, `.lab/context/auto-mode.md`, `.lab/context/auto-status.md`, and `.lab/context/auto-outcome.md`, enforce the declared terminal goal schema, make the primary gate, secondary guard, promotion condition, stop reason, and escalation reason explicit, orchestrate approved run, iterate, review, and report stages inside that contract, poll long-running work until completion or stop conditions, and write progress plus the final outcome back into `.lab/context/auto-status.md` and `.lab/context/auto-outcome.md`.
+The first visible block must be `Auto preflight`. That first visible block must list the files read and echo `Autonomy level`, `Approval status`, `Allowed stages`, `Terminal goal`, `Primary gate`, and `Secondary guard` before any execution summary or action plan.
+If the preflight block cannot be completed because any required field is missing, stale, or inconsistent, stop before execution and say exactly which field blocked arming the loop.
 When the repository workflow language is Chinese, summaries, checklist items, task labels, and progress updates should be written in Chinese unless a literal identifier must stay unchanged.
 Treat `Layer 3`, `Phase 1`, or `Table 2` as paper-scope targets. Treat `Autonomy level L3` as the execution permission level.
 Do not replace the real long-running experiment command with a short watcher such as `sleep 30`, `pgrep`, or a one-shot `metrics.json` probe. While the real experiment process is still alive, emit only a progress update and keep waiting.

package/package-assets/claude/commands/lab.md

@@ -22,7 +22,7 @@ Use the same repository artifacts and stage boundaries every time.
   Turn the approved idea into an approved dataset and benchmark package with dataset years, papers that used each dataset, source audit, download plan, classic-public versus recent-strong-public versus claim-specific benchmark roles, and explicit rationale for canonical baselines, strong historical baselines, recent strong public methods, and closest prior work.
 
 - `/lab auto ...` or `/lab-auto`
-  Run a bounded orchestration loop over approved execution stages. Use an auto-mode contract plus live auto-status to drive `run`, `iterate`, `review`, `report`, and optionally `write` without changing the frozen mission or framing. Choose an autonomy level, declare a concrete terminal goal, explicitly approve the contract before starting, and treat `.lab/context/eval-protocol.md` as the source of truth for metrics, metric glossary, source-backed comparison semantics, tables, and structured experiment-ladder rungs.
+  Run a bounded orchestration loop over approved execution stages. Use an auto-mode contract plus live auto-status to drive `run`, `iterate`, `review`, `report`, and optionally `write` without changing the frozen mission or framing. Choose an autonomy level, declare a concrete terminal goal, make the primary gate, secondary guard, promotion condition, stop reason, and escalation reason explicit, explicitly approve the contract before starting, and treat `.lab/context/eval-protocol.md` as the source of truth for metrics, metric glossary, source-backed comparison semantics, tables, and structured experiment-ladder rungs.
 
 - `/lab framing ...` or `/lab-framing`
   Lock paper-facing method name, module names, paper title, and contribution bullets by auditing current-field and adjacent-field terminology, then keep an approval gate before any section drafting.
@@ -49,6 +49,7 @@ Use the same repository artifacts and stage boundaries every time.
 ## Dispatch Rules
 
 - Always use `skills/lab/SKILL.md` as the workflow contract.
+- Use `skills/lab/references/recipes.md` as the quick path for common stage chains; do not invent new slash commands or aliases from it.
 - When the user explicitly invokes `/lab <stage> ...` or a direct `/lab-<stage>` alias, execute that stage now against the provided argument instead of only recommending another lab stage.
 - Start by giving the user a concise stage summary. Materialize managed artifacts immediately when the stage contract requires them, then report the output path and next step.
 - When ambiguity matters, ask one clarifying question at a time; when multiple paths are viable, present 2-3 approaches before converging.
@@ -69,6 +70,9 @@ Use the same repository artifacts and stage boundaries every time.
 - `L3` is the aggressive campaign level. Use it only when you explicitly want broad exploration, larger search space changes, and optional manuscript-writing work.
 - If you are unsure, choose `L2`.
 - If the request omits the level or mixes it with a paper layer, phase, or table target, `/lab auto` should stop and ask for an explicit autonomy level before arming the loop.
+- The first visible output of a real `/lab auto` run must be `Auto preflight`.
+- That first visible output must show files read plus `Autonomy level`, `Allowed stages`, `Terminal goal`, `Primary gate`, and `Secondary guard`.
+- If the preflight block cannot be completed from `.lab/context/eval-protocol.md`, `.lab/context/auto-mode.md`, `.lab/context/auto-status.md`, and `.lab/context/auto-outcome.md`, `/lab auto` should stop instead of acting like the loop is armed.
 
 - Treat `Autonomy level L1/L2/L3` as the execution privilege level, not as a paper layer, phase, or table number.
 - Treat `paper layer`, `phase`, and `table` as experiment targets. For example, `paper layer 3` or `Phase 1` should not be interpreted as `Autonomy level L3`.

package/package-assets/codex/prompts/lab-auto.md

@@ -6,7 +6,9 @@ argument-hint: autonomous campaign target
 Use the installed `lab` skill at `.codex/skills/lab/SKILL.md`.
 
 Execute the requested `/lab:auto` stage against the user's argument now. Do not only recommend another lab stage. If a blocking prerequisite is missing, say exactly what is missing and ask at most one clarifying question.
-This command runs the `/lab:auto` stage. It must read `.lab/context/eval-protocol.md`, `.lab/context/auto-mode.md`, `.lab/context/auto-status.md`, and `.lab/context/auto-outcome.md`, enforce the declared terminal goal schema, orchestrate approved run, iterate, review, and report stages inside that contract, poll long-running work until completion or stop conditions, and write progress plus the final outcome back into `.lab/context/auto-status.md` and `.lab/context/auto-outcome.md`.
+This command runs the `/lab:auto` stage. It must read `.lab/context/eval-protocol.md`, `.lab/context/auto-mode.md`, `.lab/context/auto-status.md`, and `.lab/context/auto-outcome.md`, enforce the declared terminal goal schema, make the primary gate, secondary guard, promotion condition, stop reason, and escalation reason explicit, orchestrate approved run, iterate, review, and report stages inside that contract, poll long-running work until completion or stop conditions, and write progress plus the final outcome back into `.lab/context/auto-status.md` and `.lab/context/auto-outcome.md`.
+The first visible block must be `Auto preflight`. That first visible block must list the files read and echo `Autonomy level`, `Approval status`, `Allowed stages`, `Terminal goal`, `Primary gate`, and `Secondary guard` before any execution summary or action plan.
+If the preflight block cannot be completed because any required field is missing, stale, or inconsistent, stop before execution and say exactly which field blocked arming the loop.
 When the repository workflow language is Chinese, summaries, checklist items, task labels, and progress updates should be written in Chinese unless a literal identifier must stay unchanged.
 Treat `Layer 3`, `Phase 1`, or `Table 2` as paper-scope targets. Treat `Autonomy level L3` as the execution permission level.
 Do not replace the real long-running experiment command with a short watcher such as `sleep 30`, `pgrep`, or a one-shot `metrics.json` probe. While the real experiment process is still alive, emit only a progress update and keep waiting.

package/package-assets/codex/prompts/lab.md

@@ -16,7 +16,7 @@ argument-hint: workflow question or stage choice
   Turn the approved idea into an approved dataset and benchmark package with dataset years, papers that used each dataset, source audit, download plan, classic-public versus recent-strong-public versus claim-specific benchmark roles, and explicit rationale for canonical baselines, strong historical baselines, recent strong public methods, and closest prior work.
 
 - `/lab:auto`
-  Run a bounded orchestration loop over approved execution stages. Use an auto-mode contract plus live auto-status to drive `run`, `iterate`, `review`, `report`, and optionally `write` without changing the frozen mission or framing. Choose an autonomy level, declare a concrete terminal goal, explicitly approve the contract before starting, and treat `.lab/context/eval-protocol.md` as the source of truth for metrics, metric glossary, source-backed comparison semantics, tables, and structured experiment-ladder rungs.
+  Run a bounded orchestration loop over approved execution stages. Use an auto-mode contract plus live auto-status to drive `run`, `iterate`, `review`, `report`, and optionally `write` without changing the frozen mission or framing. Choose an autonomy level, declare a concrete terminal goal, make the primary gate, secondary guard, promotion condition, stop reason, and escalation reason explicit, explicitly approve the contract before starting, and treat `.lab/context/eval-protocol.md` as the source of truth for metrics, metric glossary, source-backed comparison semantics, tables, and structured experiment-ladder rungs.
 
 - `/lab:framing`
   Lock paper-facing method name, module names, paper title, and contribution bullets by auditing current-field and adjacent-field terminology, then keep an approval gate before any section drafting.
@@ -43,6 +43,7 @@ argument-hint: workflow question or stage choice
 ## Dispatch Rules
 
 - Always use `skills/lab/SKILL.md` as the workflow contract.
+- Use `skills/lab/references/recipes.md` as the quick path for common stage chains; do not invent new slash commands or aliases from it.
 - When the user explicitly invokes `/lab:<stage>`, execute that stage now against the provided argument instead of only recommending another `/lab` stage.
 - Start by giving the user a concise stage summary. Materialize managed artifacts immediately when the stage contract requires them, then report the output path and next step.
 - When ambiguity matters, ask one clarifying question at a time; when multiple paths are viable, present 2-3 approaches before converging.
@@ -63,6 +64,9 @@ argument-hint: workflow question or stage choice
 - `L3` is the aggressive campaign level. Use it only when you explicitly want broad exploration, larger search space changes, and optional manuscript-writing work.
 - If you are unsure, choose `L2`.
 - If the request omits the level or mixes it with a paper layer, phase, or table target, `/lab:auto` should stop and ask for an explicit autonomy level before arming the loop.
+- The first visible output of a real `/lab:auto` run must be `Auto preflight`.
+- That first visible output must show files read plus `Autonomy level`, `Allowed stages`, `Terminal goal`, `Primary gate`, and `Secondary guard`.
+- If the preflight block cannot be completed from `.lab/context/eval-protocol.md`, `.lab/context/auto-mode.md`, `.lab/context/auto-status.md`, and `.lab/context/auto-outcome.md`, `/lab:auto` should stop instead of acting like the loop is armed.
 
 - Treat `Autonomy level L1/L2/L3` as the execution privilege level, not as a paper layer, phase, or table number.
 - Treat `paper layer`, `phase`, and `table` as experiment targets. For example, `paper layer 3` or `Phase 1` should not be interpreted as `Autonomy level L3`.

package/package-assets/shared/lab/.managed/templates/artifact-status.md

@@ -20,6 +20,14 @@
 - Canonical context files refreshed:
 - Evidence index anchors:
 
+## Handoff
+
+- Completed work:
+- Frozen scope:
+- Allowed next action:
+- Required read set for the next owner:
+- Accept / revise / reject boundary:
+
 ## Paper Handoff
 
 - Sections ready for `/lab:write`:

package/package-assets/shared/lab/.managed/templates/final-report.md

@@ -121,6 +121,14 @@ Preserve failed runs and rejected ideas.
 
 Describe unresolved risks and external validity limits.
 
+## Handoff
+
+- Completed work:
+- Frozen scope:
+- Allowed next action:
+- Required read set for the next owner:
+- Accept / revise / reject boundary:
+
 ## Next Steps
 
 List concrete follow-up actions.

package/package-assets/shared/lab/.managed/templates/idea.md

@@ -132,7 +132,9 @@ Suggested levels:
   - Why it survived:
 - Surviving direction 2:
   - Why it survived:
+- Why the survivors still remain:
 - Rejected directions and why:
+- Rejected Options:
 - Recommended narrowed direction:
 - Why this is stronger now:
 

package/package-assets/shared/lab/.managed/templates/review-checklist.md

@@ -21,6 +21,12 @@
 - Risk 1:
 - Risk 2:
 
+## Alternative Explanations and Boundary Risks
+
+- Strongest alternative explanation still on the table:
+- Boundary risk that could narrow the claim:
+- What evidence would rule the alternative explanation out:
+
 ## Checklist
 
 - Are the academic validity checks filled and still consistent with the actual setup?

package/package-assets/shared/lab/.managed/templates/spec.md

@@ -19,3 +19,9 @@
 - Validation run executed
 - Iteration reports generated
 - Final report generated
+
+## Failure Modes or Scenario Table
+
+- Scenario or failure mode:
+- Why it matters:
+- Mitigation or design response:

package/package-assets/shared/lab/.managed/templates/write-iteration.md

@@ -53,3 +53,11 @@
 - Continue or stop:
 - Next writing target:
 - Route back to `review` or `iterate` if needed:
+
+## Handoff
+
+- Completed work:
+- Frozen scope:
+- Allowed next action:
+- Required read set for the next owner:
+- Accept / revise / reject boundary:

package/package-assets/shared/lab/context/auto-mode.md

@@ -22,6 +22,11 @@ If `eval-protocol.md` declares structured rung entries, auto mode follows those
 - Terminal goal type:
 - Terminal goal target:
 - Required terminal artifact:
+- Primary gate:
+- Secondary guard:
+- Promotion condition:
+- Stop reason:
+- Escalation reason:
 - If the workflow language is Chinese, keep summaries, checklist items, task labels, and progress updates in Chinese.
 - Example objective: advance paper layer 3 through one bounded protocol, tests, minimal implementation, and one small run.
 

package/package-assets/shared/lab/system/core.md

@@ -36,6 +36,7 @@ For auto-mode orchestration or long-running experiment campaigns, also read:
 - `.lab/context/state.md` is a derived durable research snapshot; `.lab/context/workflow-state.md` holds live workflow state.
 - `.lab/context/summary.md` is the durable project summary; `.lab/context/session-brief.md` is the next-session startup brief.
 - `.lab/context/auto-mode.md` defines the bounded autonomous envelope; `.lab/context/auto-status.md` records live state for resume and handoff.
+- Handoff-style artifacts should use stable wording for completed work, frozen scope, allowed next action, required read set, and accept or revise or reject boundaries.
 - If the user provides a LaTeX template directory, validate it and attach it through `paper_template_root` before drafting.
 - Treat attached template directories as user-owned assets. Do not rewrite template files unless the user explicitly asks.
 - If no template is configured, use the managed default LaTeX scaffold under the deliverable paper directory.

package/package-assets/shared/skills/lab/SKILL.md

@@ -31,6 +31,7 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - Final paper output should default to LaTeX, and its manuscript language should be decided separately from the workflow language.
 - Separate sourced facts from model-generated hypotheses.
 - Preserve failed runs, failed ideas, and limitations.
+- Use `skills/lab/references/recipes.md` as the quick path for common stage chains without inventing new commands.
 
 ## Stage Contract
 
@@ -41,6 +42,7 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - For each brainstorm pass 1 candidate direction, explain what it is, why it matters, roughly how it would work, what problem it solves, and its main risk.
 - Run literature sweep 1 with closest-prior references for each candidate direction before narrowing.
 - Use brainstorm pass 2 to keep only the strongest 1-2 directions, explain why each surviving direction remains, explain what was rejected, and say why the narrowed recommendation is stronger now.
+- Name rejected options explicitly and say why the surviving directions still remain before making the final recommendation.
 - Run literature sweep 2 before making a final recommendation or novelty claim.
 - Include a user-visible literature summary that names the closest prior found, the recent strong papers found, and what existing work still does not solve before giving the final recommendation.
 - Build a literature-scoping bundle before claiming novelty. The default target is 20 relevant sources unless the field is genuinely too narrow and that exception is written down.
@@ -108,7 +110,10 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - Use this stage to orchestrate approved execution stages with bounded autonomy.
 - Read `.lab/config/workflow.json`, `.lab/context/mission.md`, `.lab/context/state.md`, `.lab/context/workflow-state.md`, `.lab/context/decisions.md`, `.lab/context/data-decisions.md`, `.lab/context/evidence-index.md`, `.lab/context/terminology-lock.md`, `.lab/context/auto-mode.md`, and `.lab/context/auto-status.md` before acting.
 - Treat `.lab/context/auto-mode.md` as the control contract and `.lab/context/auto-status.md` as the live state file.
+- Require `.lab/context/auto-mode.md` to expose `Primary gate`, `Secondary guard`, `Promotion condition`, `Stop reason`, and `Escalation reason` before execution.
 - Require `Autonomy level` and `Approval status` in `.lab/context/auto-mode.md` before execution.
+- Start every `/lab:auto` run with a visible `Auto preflight` summary that reports files read plus `Autonomy level`, `Approval status`, `Allowed stages`, `Terminal goal`, `Primary gate`, and `Secondary guard`.
+- If any required preflight field is missing or inconsistent, stop before any loop action. Do not present a fake auto summary as if the loop were armed.
 - Treat `L1` as safe-run validation, `L2` as bounded iteration, and `L3` as aggressive campaign mode.
 - Surface the level guide every time `/lab:auto` starts, and make the detailed guide mandatory when the user omits the level or mixes it with a paper layer, phase, or table target.
 - Reuse `/lab:run`, `/lab:iterate`, `/lab:review`, `/lab:report`, and optional `/lab:write` instead of inventing a second workflow.
@@ -126,6 +131,7 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - Use `.lab/changes/<change-id>/` as the canonical lab change directory.
 - Convert the approved idea into lab change artifacts using `.lab/.managed/templates/proposal.md`, `.lab/.managed/templates/design.md`, `.lab/.managed/templates/spec.md`, and `.lab/.managed/templates/tasks.md`.
 - When the approved idea involves human-subject evaluation, use this stage to freeze the human-subject experiment design: participant recruitment, sample-size rationale, condition design, assignment, measurement, and ethics or debrief details.
+- When the design depends on interaction, safety, protocol, or human-subject boundaries, add a failure-mode or scenario table instead of leaving those cases implicit.
 - Update `.lab/context/decisions.md` after freezing the spec, then refresh derived views.
 - Do not skip task definition.
 
@@ -168,6 +174,7 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - Read `.lab/context/mission.md`, `.lab/context/decisions.md`, `.lab/context/evidence-index.md`, and `.lab/context/data-decisions.md` before reviewing.
 - Start with a concise summary of what is being reviewed and the top review question.
 - Prioritize methodology, fairness, benchmark representativeness, comparison-category coverage, leakage, statistics, ablations, and claim discipline.
+- Surface the strongest alternative explanation and any boundary risk that should narrow the claim.
 - Output findings first, then fatal flaws, then fix priority, then residual risks.
 - Use `.lab/.managed/templates/review-checklist.md`.
 - Write durable review conclusions back to `.lab/context/decisions.md`, `.lab/context/evidence-index.md`, or `.lab/context/open-questions.md` when they affect later stages. Do not use `.lab/context/state.md` as a primary write target.
@@ -238,6 +245,7 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 ## References
 
 - Workflow summary: `.codex/skills/lab/references/workflow.md` or `.claude/skills/lab/references/workflow.md`
+- Stage recipes: `skills/lab/references/recipes.md`
 - Brainstorming integration: `.codex/skills/lab/references/brainstorming-integration.md` or `.claude/skills/lab/references/brainstorming-integration.md`
 - Idea stage guide: `.codex/skills/lab/stages/idea.md` or `.claude/skills/lab/stages/idea.md`
 - Data stage guide: `.codex/skills/lab/stages/data.md` or `.claude/skills/lab/stages/data.md`

package/package-assets/shared/skills/lab/references/recipes.md

@@ -0,0 +1,38 @@
+# /lab Stage Recipes
+
+Use this file as a quick routing map for common stage chains. It does not add new commands or replace stage contracts.
+
+## idea -> data -> spec
+
+Use this path when the problem is still being selected or narrowed.
+
+- Start at `/lab:idea` to compare candidate directions, run literature sweeps, and end with an approval gate.
+- Move to `/lab:data` after the idea is approved and the benchmark package needs to be frozen.
+- Move to `/lab:spec` after the idea and data package are approved and the change must be turned into an actionable design.
+
+## run -> iterate -> review -> report
+
+Use this path when the mission is already approved and the work is now evidence generation.
+
+- Start at `/lab:run` for the smallest meaningful validation.
+- Move to `/lab:iterate` when the campaign needs bounded rounds against a frozen mission.
+- Move to `/lab:review` when methodology, fairness, leakage, or claim discipline needs reviewer-style scrutiny.
+- Move to `/lab:report` when the validated evidence is stable enough for collaborator-facing synthesis.
+
+## framing -> write -> review
+
+Use this path when report artifacts are stable and the paper narrative is being finalized.
+
+- Start at `/lab:framing` to lock paper-facing names, title direction, and contribution wording.
+- Move to `/lab:write` for section-by-section manuscript drafting against the approved framing and report evidence.
+- Revisit `/lab:review` when a draft section or paper-level claim set needs reviewer-mode critique before finalization.
+
+## `/lab:auto` by approved scope
+
+Use `/lab:auto` only after the upstream stage decisions are already approved.
+
+- `L1`: one bounded validation cycle over approved execution stages
+- `L2`: default bounded experiment iteration inside a frozen core
+- `L3`: broader campaign that may include writing only when framing is already approved
+
+`/lab:auto` is an orchestration layer over approved stage chains, not a replacement for `idea`, `data`, `framing`, or `spec`.

package/package-assets/shared/skills/lab/references/workflow.md

@@ -1,5 +1,7 @@
 # /lab Workflow Reference
 
+For common stage chains, see the quick-path guide in `skills/lab/references/recipes.md`.
+
 ## Frozen Mission
 
 Within `/lab:iterate`, the campaign target stays fixed across rounds.

package/package-assets/shared/skills/lab/stages/auto.md

@@ -2,6 +2,7 @@
 
 ## Required Output
 
+- visible `Auto preflight` summary
 - Auto Mode Contract
 - approved auto-mode contract
 - live auto-status state
@@ -46,12 +47,19 @@
 - Treat `Academic Validity Checks` and `Integrity self-check` as mandatory automation gates. Auto mode should not proceed, promote, or declare success while those fields are missing, stale, or contradicted by the current rung.
 - Treat `Sanity and Alternative-Explanation Checks` as the anomaly gate for automation. When a rung yields all-null outputs, suspiciously identical runs, no-op deltas, or impl/result mismatches, pause promotion logic until implementation reality checks, alternative explanations, and at least one cross-check are recorded.
 - Treat paper-template selection as an explicit write-time gate, not as a silent fallback, when the loop is about to create `.tex` deliverables for the first time.
+- Treat `.lab/context/auto-mode.md` as a visible control plane. The contract should make the primary gate, secondary guard, promotion condition, stop reason, and escalation reason explicit before execution starts.
 - The contract must declare `Autonomy level` and `Approval status`, and execution starts only when approval is explicitly set to `approved`.
 - The contract must also declare a concrete terminal goal:
   - `rounds`
   - `metric-threshold`
   - `task-completion`
 - The contract must provide both `Terminal goal target` and `Required terminal artifact`.
+- Use these control-plane terms consistently:
+  - `Primary gate`: the main pass condition that decides whether the current rung or loop objective is satisfied
+  - `Secondary guard`: the safety or anomaly check that can block a pass even when the primary gate looks good
+  - `Promotion condition`: the explicit rule that allows exploratory work to be promoted into the primary package
+  - `Stop reason`: the concrete boundary that ends the current loop without promotion
+  - `Escalation reason`: the concrete condition that forces human review or a narrower next step
 - Recommended level meanings:
   - `L1`: safe run validation over `run`, `review`, and `report`
   - `L2`: bounded iteration over `run`, `iterate`, `review`, and `report`
@@ -64,7 +72,17 @@
 - Keep a poll-based waiting loop instead of sleeping blindly.
 - Do not treat a short watcher such as `sleep 30`, a one-shot `pgrep`, or a single `metrics.json` probe as the rung command when the real experiment is still running.
 - Bind each rung to the real long-running command or process that owns the experiment result.
+- Start every real `/lab:auto` run with a visible `Auto preflight` block before any execution summary or action plan. That first visible output should list:
+  - files read
+  - `Autonomy level`
+  - `Approval status`
+  - `Allowed stages`
+  - `Terminal goal`
+  - `Primary gate`
+  - `Secondary guard`
+- If any of those preflight fields are missing, stale, or inconsistent, stop before execution and report the blocking field directly.
 - Always write a canonical `.lab/context/auto-outcome.md` when the run completes, stops, or fails.
+- Keep handoff wording stable across auto outcomes and downstream report or write handoffs: record completed work, frozen scope, allowed next action, required read set for the next owner, and the accept or revise or reject boundary.
 - When the evaluation protocol declares structured ladder rungs, execute them as a foreground rung state machine:
   - each rung must declare `Stage`, `Goal`, `Command`, `Watch`, `Gate`, `On pass`, `On fail`, and `On stop`
   - keep the session alive while the current rung is running
@@ -85,15 +103,16 @@
 ## Minimum Procedure
 
 1. Validate the auto-mode contract
-2. Confirm the approved autonomy level matches the requested stage envelope
-3. Set or refresh auto-status
-4. Choose the next allowed `/lab` stage or structured ladder rung
-5. Launch the bounded action
-6. Poll for process completion, checkpoint movement, or summary generation while keeping the session alive
-7. Evaluate the declared rung gate and transition to the next rung when structured ladder mode is active
-8. Evaluate the declared terminal goal semantics at the correct boundary
-9. Evaluate stop, success, and promotion checks at the correct boundary
-10. Write auto-outcome and decide continue, promote, stop, or escalate
+2. Emit the visible `Auto preflight` summary
+3. Confirm the approved autonomy level matches the requested stage envelope
+4. Set or refresh auto-status
+5. Choose the next allowed `/lab` stage or structured ladder rung
+6. Launch the bounded action
+7. Poll for process completion, checkpoint movement, or summary generation while keeping the session alive
+8. Evaluate the declared rung gate and transition to the next rung when structured ladder mode is active
+9. Evaluate the declared terminal goal semantics at the correct boundary
+10. Evaluate the primary gate, secondary guard, promotion condition, stop reason, and escalation reason at the correct boundary
+11. Write auto-outcome and decide continue, promote, stop, or escalate
 
 ## Interaction Contract
 
@@ -105,6 +124,7 @@
 - If you are unsure, choose `L2`.
 
 - Start with a concise summary of the objective, the frozen core, and the next automatic stage.
+- The first visible output must be `Auto preflight`, not a result summary or stage recommendation.
 - Always surface the level guide before execution.
 - If the contract is incomplete, ask one clarifying question at a time.
 - If multiple next actions are credible, present 2-3 bounded options with trade-offs before arming a long run.

package/package-assets/shared/skills/lab/stages/idea.md

@@ -23,6 +23,8 @@
 - literature sweep 1 with 3-5 closest-prior references per direction
 - brainstorm pass 2 that narrows to 1-2 surviving directions
 - each brainstorm pass 2 surviving direction explained with why it survived, plus rejected directions and why, and why the narrowed recommendation is stronger now
+- rejected options stated explicitly, not only implied
+- why the survivors remain stated explicitly, not only implied
 - literature sweep 2 that expands the surviving directions into the full source bundle
 - literature summary for recommendation with closest prior found, recent strong papers found, and what existing work still does not solve
 - literature scoping bundle with a default target of 20 sources, or an explicit explanation for a smaller scoped field
@@ -126,6 +128,8 @@
 - Give brainstorm pass 1 at least three candidate directions, and explain each one with what it is, why it matters, rough how it would work, what problem it solves, and its main risk.
 - Use literature sweep 1 to test candidate directions against real papers before narrowing them.
 - Use brainstorm pass 2 to explain what survived, why it survived, what was rejected, why it was rejected, and why the narrowed recommendation is stronger now.
+- Name rejected options explicitly and keep a short explanation of why each one lost.
+- Name why the survivors remain explicitly so later stages inherit the narrowed logic instead of only the final winner label.
 - Use literature sweep 2 to support the final recommendation with real references across the required buckets.
 - Add a short literature summary for recommendation so the final output shows the closest prior found, the recent strong papers found, and what existing work still does not solve.
 - Compare against existing methods explicitly, not by vague novelty language.

package/package-assets/shared/skills/lab/stages/report.md

@@ -23,6 +23,7 @@
 - limitations
 - next steps
 - artifact status kept separate from validated findings
+- stable handoff section that says what is complete, what is frozen, what the next owner may do, what they must read, and what counts as accept, revise, or reject
 
 ## Context Read Set
 
@@ -71,6 +72,7 @@
 - If the existing `report.md` or `main-tables.md` is missing required collaborator-facing sections from the managed templates, treat that as a report deficiency. A rerun must repair the missing sections instead of declaring "no content change" or treating the rerun as a no-op.
 - After drafting or rerunning the report, run `.lab/.managed/scripts/validate_collaborator_report.py --report <deliverables_root>/report.md --main-tables <deliverables_root>/main-tables.md`. If it fails, keep editing until it passes; do not stop at a no-op audit rerun.
 - Do not mix workflow deliverable status, rerun ids, or manuscript skeleton status into validated scientific findings; keep those in `<deliverables_root>/artifact-status.md`.
+- Keep handoff wording stable across `report.md` and `<deliverables_root>/artifact-status.md`: record completed work, frozen scope, allowed next action, the required read set for the next owner, and the accept or revise or reject boundary.
 - Write durable report-level conclusions into canonical context such as `.lab/context/mission.md`, `.lab/context/eval-protocol.md`, `.lab/context/decisions.md`, and `.lab/context/evidence-index.md`, then refresh the derived `state.md` snapshot. Write live reporting progress or immediate handoff actions into `.lab/context/workflow-state.md`.
 - If `.lab/config/workflow.json` sets the workflow language to Chinese, write `report.md` and `<deliverables_root>/main-tables.md` in Chinese unless a file path, code identifier, or literal metric name must remain unchanged.
 - Prefer conservative interpretation over marketing language.

package/package-assets/shared/skills/lab/stages/review.md

@@ -38,6 +38,8 @@
 - unsupported causal or statistical claims
 - missing ablations
 - irreproducible reporting
+- unresolved alternative explanations
+- boundary risks that should narrow the claim even if the implementation is correct
 
 ## Output Style
 
@@ -47,6 +49,7 @@
 - fix priority stated clearly
 - evidence-linked critique
 - explicit residual risks
+- explicit alternative explanations and boundary risks
 
 ## Interaction Contract
 
@@ -54,3 +57,4 @@
 - Ask one clarifying question at a time only if review scope ambiguity would change findings or severity.
 - If there are multiple legitimate review framings, present 2-3 approaches with trade-offs and recommend the strictest useful framing.
 - Do not use brainstorming to soften critique; once scope is clear, stay in reviewer mode and deliver findings directly.
+- Call out the strongest remaining alternative explanation and the strongest boundary risk when either one could materially narrow the claim.

package/package-assets/shared/skills/lab/stages/spec.md

@@ -38,6 +38,7 @@
 - Carry the approved dataset package, source choices, and benchmark mix into the change.
 - Preserve evaluation boundaries from the idea stage.
 - If the approved idea includes human-subject evaluation, convert the rough evaluation sketch into an explicit human-subject experiment design instead of leaving recruitment or protocol details implicit.
+- When interaction, safety, protocol, or human-subject risk matters, add a failure-mode or scenario table instead of leaving those edge cases implicit.
 - Translate risks into concrete tasks when possible.
 - Make task granularity small enough that `/lab:run` and `/lab:iterate` can execute predictably.
 - Use one lab-native change directory per approved idea instead of scattering spec artifacts.
@@ -48,6 +49,7 @@
 - If change decomposition or validation strategy is materially ambiguous, ask one clarifying question at a time.
 - If there are multiple viable ways to structure the lab change directory, present 2-3 approaches with trade-offs and a recommendation before freezing the change.
 - Keep an approval gate before locking a change structure that will drive `/lab:run` and `/lab:iterate`.
+- Make failure modes or scenario coverage explicit when the design depends on interaction boundaries, hidden assumptions, safety constraints, or human-subject protocol choices.
 
 ## Minimum Task Coverage
 
@@ -64,6 +66,7 @@
 - assignment or randomization protocol
 - measurement or survey plan
 - ethics, consent, and debrief plan when applicable
+- failure-mode or scenario table when the approved idea depends on interaction, safety, protocol, or human-subject boundaries
 - artifact creation
 - validation run
 - evaluation normalization

package/package-assets/shared/skills/lab/stages/write.md

@@ -146,6 +146,7 @@ Run these on every round:
 - When a round introduces or revises key terms, include a compact terminology note in the user-facing round summary and record the terminology-clarity self-check in the write-iteration artifact.
 - When `workflow_language` and `paper_language` differ, record the final manuscript language choice in the write-iteration artifact with the workflow language, paper language, finalization decision, and why that decision was chosen.
 - When `workflow_language` and `paper_language` differ, also record the persisted workflow-language deliverable path in the write-iteration artifact.
+- Keep the handoff wording in the write-iteration artifact explicit and stable: say what is completed, what scope is frozen for the next round, what the next owner is allowed to do, what they must read first, and what counts as accept, revise, or reject.
 - Return paragraph-level roles for the revised prose when drafting.
 - Run the five-dimension self-review checklist before accepting a round.
 - Run reviewer-style checks after every round.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "superlab",
-  "version": "0.1.40",
+  "version": "0.1.42",
   "description": "Strict /lab research workflow installer for Codex and Claude",
   "keywords": [
     "codex",