superlab 0.1.15 → 0.1.17

package/README.md

@@ -45,7 +45,7 @@ This writes:
 
 - `.codex/prompts/lab-*.md`
 - `.codex/skills/lab/`
-- `.claude/commands/lab/*.md`
+- `.claude/commands/lab*.md`
 - `.claude/skills/lab/`
 - `AGENTS.md`
 - `CLAUDE.md`
@@ -250,6 +250,11 @@ Stages should follow that file rather than guess language locally.
 
 ## Command Set
 
+Codex and Claude use different slash-entry syntax:
+
+- Codex: `/lab:idea`, `/lab:auto`, `/lab:write`
+- Claude Code: `/lab idea ...` or `/lab-idea`; `/lab auto ...` or `/lab-auto`; `/lab write ...` or `/lab-write`
+
 - `/lab:idea` researches an idea, critiques it, and writes the initial research framing.
 - `/lab:data` turns the approved idea into a dataset and benchmark package with years, paper usage, source audit, download plans, explicit benchmark-role rationale for classic-public, recent-strong-public, and claim-specific benchmarks, and explicit comparison rationale for canonical baselines, strong historical baselines, recent strong public methods, and closest prior work.
 - `/lab:framing` locks paper-facing method names, module names, titles, and contribution wording before drafting.
@@ -281,15 +286,15 @@ See the source command docs in [commands/codex/lab.md](/Users/zhouhao119/coding/
 
 ## Typical Flow
 
-1. Run `/lab:idea` to produce the initial idea artifact.
-2. Run `/lab:data` to lock the approved dataset package, source hierarchy, benchmark-class coverage, and the rationale for each comparison-method class.
-3. Run `/lab:spec` to convert the idea into one self-contained lab change directory.
-4. Run `/lab:run` to verify the experiment path and register the first run.
-5. Run `/lab:iterate` to improve against fixed targets using bounded iterations.
-6. Run `/lab:review` whenever you need reviewer-grade critique.
-7. Run `/lab:report` to synthesize the final report.
-8. Run `/lab:framing` to lock naming, title, and contribution wording.
-9. Run `/lab:write` to draft paper sections from stable report evidence.
+1. In Codex, run `/lab:idea`; in Claude Code, run `/lab idea ...` or `/lab-idea`.
+2. In Codex, run `/lab:data`; in Claude Code, run `/lab data ...` or `/lab-data`.
+3. In Codex, run `/lab:spec`; in Claude Code, run `/lab spec ...` or `/lab-spec`.
+4. In Codex, run `/lab:run`; in Claude Code, run `/lab run ...` or `/lab-run`.
+5. In Codex, run `/lab:iterate`; in Claude Code, run `/lab iterate ...` or `/lab-iterate`.
+6. In Codex, run `/lab:review`; in Claude Code, run `/lab review ...` or `/lab-review`.
+7. In Codex, run `/lab:report`; in Claude Code, run `/lab report ...` or `/lab-report`.
+8. In Codex, run `/lab:framing`; in Claude Code, run `/lab framing ...` or `/lab-framing`.
+9. In Codex, run `/lab:write`; in Claude Code, run `/lab write ...` or `/lab-write`.
 
 `/lab:write` writes final manuscript output under the configured `deliverables_root` (default: `docs/research`):
 

package/README.zh-CN.md

@@ -43,7 +43,7 @@ npx github:zhouhaoUCAS/superlab init
 
 - `.codex/prompts/lab-*.md`
 - `.codex/skills/lab/`
-- `.claude/commands/lab/*.md`
+- `.claude/commands/lab*.md`
 - `.claude/skills/lab/`
 - `AGENTS.md`
 - `CLAUDE.md`
@@ -248,6 +248,11 @@ superlab init --lang en
 
 ## 命令集合
 
+Codex 和 Claude 的命令入口不一样：
+
+- Codex：`/lab:idea`、`/lab:auto`、`/lab:write`
+- Claude Code：`/lab idea ...` 或 `/lab-idea`；`/lab auto ...` 或 `/lab-auto`；`/lab write ...` 或 `/lab-write`
+
 - `/lab:idea` 调研 idea、文献、数据集、指标和 baseline，并输出初始方案。
 - `/lab:data` 把已批准的 idea 收敛成数据集与 benchmark 方案，要求记录年份、使用论文、来源审计、下载计划，并明确 classic-public、recent-strong-public、claim-specific 三类 benchmark 的纳入理由，以及 canonical baselines、strong historical baselines、recent strong public methods、closest prior work 四类对比方法的纳入理由。
 - `/lab:framing` 在正式写作前收紧方法名、模块名、论文题目和 contribution wording。
@@ -266,15 +271,15 @@ superlab init --lang en
 ## 使用流程
 
 1. 在目标项目执行 `superlab init`。
-2. 在 Codex 或 Claude 中直接调用 `/lab:idea`。
-3. 经确认后执行 `/lab:data`，锁定数据集、下载来源、benchmark 类别覆盖，以及各类对比方法的纳入理由。
-4. 再执行 `/lab:spec`。
-5. 用 `/lab:run` 打通最小实验链路。
-6. 用 `/lab:iterate` 进行多轮迭代。
-7. 在关键节点运行 `/lab:review`。
-8. 最后用 `/lab:report` 产出总报告。
-9. 用 `/lab:framing` 收紧题目、命名和 contribution wording。
-10. 用 `/lab:write` 把稳定结果写成论文各 section。
+2. 在 Codex 中调用 `/lab:idea`；在 Claude Code 中调用 `/lab idea ...` 或 `/lab-idea`。
+3. 经确认后，在 Codex 中执行 `/lab:data`，或在 Claude Code 中执行 `/lab data ...` / `/lab-data`，锁定数据集、下载来源、benchmark 类别覆盖，以及各类对比方法的纳入理由。
+4. 再执行 `/lab:spec`，或在 Claude Code 中执行 `/lab spec ...` / `/lab-spec`。
+5. 用 `/lab:run`，或在 Claude Code 中用 `/lab run ...` / `/lab-run` 打通最小实验链路。
+6. 用 `/lab:iterate`，或在 Claude Code 中用 `/lab iterate ...` / `/lab-iterate` 进行多轮迭代。
+7. 在关键节点运行 `/lab:review`，或在 Claude Code 中运行 `/lab review ...` / `/lab-review`。
+8. 最后用 `/lab:report`，或在 Claude Code 中用 `/lab report ...` / `/lab-report` 产出总报告。
+9. 用 `/lab:framing`，或在 Claude Code 中用 `/lab framing ...` / `/lab-framing` 收紧题目、命名和 contribution wording。
+10. 用 `/lab:write`，或在 Claude Code 中用 `/lab write ...` / `/lab-write` 把稳定结果写成论文各 section。
 
 `/lab:write` 会把最终可交付物写到 `deliverables_root` 指定的目录，默认是 `docs/research`：
 

package/lib/i18n.cjs

@@ -10,12 +10,11 @@ ${body}
 `;
 }
 
-function claudeCommand(name, description, tags, body) {
+function claudeCommand(name, description, argumentHint, body) {
   return `---
 name: "${name}"
 description: ${description}
-category: 工作流
-tags: [${tags}]
+argument-hint: ${argumentHint}
 ---
 
 ${body}
@@ -56,60 +55,60 @@ const ZH_CONTENT = {
   [path.join(".codex", "prompts", "lab-report.md")]: codexPrompt(
     "基于验证后的迭代工件生成最终报告",
     "report context",
-    "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:report`，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:report` 阶段。它必须汇总标准化摘要、保留失败尝试和局限，并生成最终实验报告。"
+    "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:report`，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:report` 阶段。它必须生成给用户直接阅读的最终实验报告和受管的 `main-tables.md`，明确写出主指标、次级指标和必要终局证据，并用白话解释这些指标分别衡量什么、哪些只是健康度或支持性指标、以及每张主表到底证明了什么和没证明什么。"
   ),
   [path.join(".codex", "prompts", "lab-write.md")]: codexPrompt(
     "把验证过的研究工件转成论文 section，并按小步方式修订",
     "section or writing target",
     "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:write`，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:write` 阶段。它必须先有来自 `/lab:framing` 的已批准 framing artifact，再读取 `.codex/skills/lab/references/paper-writing/` 下与当前 section 对应的参考文件；如果当前是 `abstract`、`introduction` 或 `method`，还必须继续读取 `.codex/skills/lab/references/paper-writing/examples/index.md`、对应的 examples index，以及 1-2 个具体 example 文件。然后结合 `paper-review.md` 与 `does-my-writing-flow-source.md`，先写 mini-outline，再只修改一个 section。第一次进入会产出论文 `.tex` 的路径时，如果 `paper_template_root` 为空，必须先问一次：继续使用默认 LaTeX scaffold，还是先接入模板目录。"
   ),
-  [path.join(".claude", "commands", "lab", "idea.md")]: claudeCommand(
-    "LAB: Idea",
+  [path.join(".claude", "commands", "lab-idea.md")]: claudeCommand(
+    "lab-idea",
     "在进入规格前调研并打磨论文或实验想法",
-    "workflow, research, idea",
-    "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 `/lab:idea`，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:idea` 阶段。它必须先用清晰简洁的话定义问题与失败场景，说明现有方法哪里不够、我们的想法为何更好，再做 idea classification、contribution category、breakthrough level 的归类，并收束出至少三个一眼就有意义的点，最后保留进入 `/lab:spec` 前的 approval gate。"
+    "idea 或 research problem",
+    "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 `idea` 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 `idea` 阶段。它必须先用清晰简洁的话定义问题与失败场景，说明现有方法哪里不够、我们的想法为何更好，再做 idea classification、contribution category、breakthrough level 的归类，并收束出至少三个一眼就有意义的点，最后保留进入 `spec` 前的 approval gate。"
   ),
-  [path.join(".claude", "commands", "lab", "framing.md")]: claudeCommand(
-    "LAB: Framing",
+  [path.join(".claude", "commands", "lab-framing.md")]: claudeCommand(
+    "lab-framing",
     "在写作前收紧方法名、模块名、论文题目和 contribution framing",
-    "workflow, research, framing",
-    "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 `/lab:framing`，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:framing` 阶段。它必须围绕 paper-facing framing，收紧方法名、模块名、题目和 contribution bullets，审查当前领域与相邻领域的术语是否 canonical、对象是否明确，并在进入 `/lab:write` 前保留 approval gate。"
+    "naming、title 或 contribution framing target",
+    "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 `framing` 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 `framing` 阶段。它必须围绕 paper-facing framing，收紧方法名、模块名、题目和 contribution bullets，审查当前领域与相邻领域的术语是否 canonical、对象是否明确，并在进入 `write` 前保留 approval gate。"
   ),
-  [path.join(".claude", "commands", "lab", "spec.md")]: claudeCommand(
-    "LAB: Spec",
+  [path.join(".claude", "commands", "lab-spec.md")]: claudeCommand(
+    "lab-spec",
     "把已批准的 idea 转成统一的 lab change 目录",
-    "workflow, research, spec",
-    "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 `/lab:spec`，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:spec` 阶段。它必须围绕一个 change id，在 `.lab/changes/<change-id>/` 下生成 proposal/design/spec/tasks，并让这个 change 成为后续 run 和 iterate 的统一入口。"
+    "approved idea context",
+    "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 `spec` 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 `spec` 阶段。它必须围绕一个 change id，在 `.lab/changes/<change-id>/` 下生成 proposal/design/spec/tasks，并让这个 change 成为后续 run 和 iterate 的统一入口。"
   ),
-  [path.join(".claude", "commands", "lab", "run.md")]: claudeCommand(
-    "LAB: Run",
+  [path.join(".claude", "commands", "lab-run.md")]: claudeCommand(
+    "lab-run",
     "执行最小可行实验并标准化输出",
-    "workflow, research, run",
-    "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 `/lab:run`，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:run` 阶段。它必须从最小实验开始，登记 run，生成标准化评估摘要，并验证结果格式。"
+    "run context",
+    "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 `run` 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 `run` 阶段。它必须从最小实验开始，登记 run，生成标准化评估摘要，并验证结果格式。"
   ),
-  [path.join(".claude", "commands", "lab", "iterate.md")]: claudeCommand(
-    "LAB: Iterate",
+  [path.join(".claude", "commands", "lab-iterate.md")]: claudeCommand(
+    "lab-iterate",
     "在固定成功标准下执行有边界的实验迭代",
-    "workflow, research, iterate",
-    "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 `/lab:iterate`，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:iterate` 阶段。它必须冻结 mission、声明 completion_promise、只做小步改动、每轮生成评估和 iteration report；若风险连续两轮升高则切 diagnostic mode，并在失败结束时记录 blockers 与 next actions。"
+    "iteration mission",
+    "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 `iterate` 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 `iterate` 阶段。它必须冻结 mission、声明 completion_promise、只做小步改动、每轮生成评估和 iteration report；若风险连续两轮升高则切 diagnostic mode，并在失败结束时记录 blockers 与 next actions。"
   ),
-  [path.join(".claude", "commands", "lab", "review.md")]: claudeCommand(
-    "LAB: Review",
+  [path.join(".claude", "commands", "lab-review.md")]: claudeCommand(
+    "lab-review",
     "以审稿人模式审查研究方案或结果",
-    "workflow, research, review",
-    "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 `/lab:review`，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:review` 阶段。它必须先给简洁摘要，再按 findings -> fatal flaws -> fix priority -> residual risks 的顺序输出，优先检查方法学漏洞、对照公平性、数据泄漏、统计不足和 unsupported claims。"
+    "artifact or result to review",
+    "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 `review` 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 `review` 阶段。它必须先给简洁摘要，再按 findings -> fatal flaws -> fix priority -> residual risks 的顺序输出，优先检查方法学漏洞、对照公平性、数据泄漏、统计不足和 unsupported claims。"
   ),
-  [path.join(".claude", "commands", "lab", "report.md")]: claudeCommand(
-    "LAB: Report",
+  [path.join(".claude", "commands", "lab-report.md")]: claudeCommand(
+    "lab-report",
     "基于验证后的迭代工件生成最终报告",
-    "workflow, research, report",
-    "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 `/lab:report`，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:report` 阶段。它必须汇总标准化摘要、保留失败尝试和局限，并生成最终实验报告。"
+    "report context",
+    "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 `report` 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 `report` 阶段。它必须生成给用户直接阅读的最终实验报告和受管的 `main-tables.md`，明确写出主指标、次级指标和必要终局证据，并用白话解释这些指标分别衡量什么、哪些只是健康度或支持性指标、以及每张主表到底证明了什么和没证明什么。"
   ),
-  [path.join(".claude", "commands", "lab", "write.md")]: claudeCommand(
-    "LAB: Write",
+  [path.join(".claude", "commands", "lab-write.md")]: claudeCommand(
+    "lab-write",
     "把验证过的研究工件转成论文 section，并按小步方式修订",
-    "workflow, research, writing",
-    "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 `/lab:write`，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:write` 阶段。它必须先有来自 `/lab:framing` 的已批准 framing artifact，再读取 `.claude/skills/lab/references/paper-writing/` 下与当前 section 对应的参考文件；如果当前是 `abstract`、`introduction` 或 `method`，还必须继续读取 `.claude/skills/lab/references/paper-writing/examples/index.md`、对应的 examples index，以及 1-2 个具体 example 文件。然后结合 `paper-review.md` 与 `does-my-writing-flow-source.md`，先写 mini-outline，再只修改一个 section。第一次进入会产出论文 `.tex` 的路径时，如果 `paper_template_root` 为空，必须先问一次：继续使用默认 LaTeX scaffold，还是先接入模板目录。"
+    "section or writing target",
+    "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 `write` 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 `write` 阶段。它必须先有来自 `framing` 阶段的已批准 framing artifact，再读取 `.claude/skills/lab/references/paper-writing/` 下与当前 section 对应的参考文件；如果当前是 `abstract`、`introduction` 或 `method`，还必须继续读取 `.claude/skills/lab/references/paper-writing/examples/index.md`、对应的 examples index，以及 1-2 个具体 example 文件。然后结合 `paper-review.md` 与 `does-my-writing-flow-source.md`，先写 mini-outline，再只修改一个 section。第一次进入会产出论文 `.tex` 的路径时，如果 `paper_template_root` 为空，必须先问一次：继续使用默认 LaTeX scaffold，还是先接入模板目录。"
   ),
 };
 
@@ -290,11 +289,14 @@ const ZH_SKILL_FILES = {
 
 ## 必要输出
 
+- 给用户看的总结
 - 方法概述
 - 选定指标摘要
+- 指标白话释义
 - 实验设置
 - 已验证主结果
 - 位于 \`<deliverables_root>/main-tables.md\` 的受管主表工件
+- 怎么看主表的阅读指引
 - 消融
 - 失败尝试
 - 局限性
@@ -320,6 +322,8 @@ const ZH_SKILL_FILES = {
 - 主表结构、gate 和最终结果 framing 必须对齐已批准的评估协议。
 - 不要凭记忆重述指标定义、baseline 行为或对比方法实现；直接引用评估协议里记录的来源。
 - 必须把已批准的主指标、次级指标和必要终局证据明确写进 \`report.md\` 与受管的 \`main-tables.md\`。
+- 必须用白话解释选定的主指标和次级指标：每个指标在衡量什么、越高还是越低更好、它是主结果指标还是健康度/支持性指标。
+- 如果出现 coverage、completeness、confidence 或类似健康度指标，必须明确说明这类指标回答的是“实验是否跑稳、证据是否完整”，而不是主要科学效应本身。
 - 如果报告依赖了对原始指标或原始实现的偏差，必须明确写出这个偏差。
 - 如果 workflow language 是中文，\`report.md\` 和 \`<deliverables_root>/main-tables.md\` 也应使用中文，除非文件路径、代码标识符或字面指标名必须保持原样。
 - 解释优先保守，不要写成营销文案。
@@ -328,6 +332,8 @@ const ZH_SKILL_FILES = {
 ## 交互约束
 
 - 开始前先简洁说明：campaign outcome、选定的主指标和次级指标、最强已支撑 claim、最大的报告风险。
+- 当该阶段由 \`/lab:auto\` 进入时，要主动给出用户可读的白话总结，不要等用户再追问“这些指标是什么意思”或“这些表怎么看”。
+- 把 \`report.md\` 当作给用户看的工件，而不是内部 dump。术语第一次出现时就解释；先讲结论，再讲术语。
 - 如果某个未决前提会改变报告解释，一次只问一个问题。
 - 如果存在多种报告 framing，先给 2-3 个方案、trade-offs 和推荐项，优先最忠于证据的 framing。
 - 如果某种 framing 会实质影响后续论文 claim，要保留 approval gate。
@@ -685,10 +691,12 @@ const ZH_SKILL_FILES = {
   [path.join(".lab", ".managed", "templates", "final-report.md")]:
 `# 最终报告
 
-## 目标
+## 给用户看的总结
 
-- 本轮研究目标：
-- 是否达标：
+- 一句话结论：
+- 已经被验证的内容：
+- 还没有被证明的内容：
+- 当前最大报告风险：
 
 ## 选定指标
 
@@ -696,6 +704,19 @@ const ZH_SKILL_FILES = {
 - 次级指标：
 - 必要终局证据：
 
+## 指标白话释义
+
+- 主指标在衡量什么：
+- 次级指标在衡量什么：
+- 健康度/支持性指标在衡量什么，为什么它们不是主结论：
+
+## 怎么看主表
+
+- Table 1 负责回答什么：
+- Table 2 负责回答什么：
+- Table 3 负责回答什么：
+- Table 4 负责回答什么：
+
 ## 主表工件
 
 - 受管主表路径：\`<deliverables_root>/main-tables.md\`
@@ -720,18 +741,37 @@ const ZH_SKILL_FILES = {
   [path.join(".lab", ".managed", "templates", "main-tables.md")]:
 `# 主表工件
 
+## 给用户看的总结
+
+- 用户可直接复述的结论：
+- 这些主表证明了什么：
+- 这些主表还不能证明什么：
+
 ## 选定指标
 
 - 主指标：
 - 次级指标：
 - 必要终局证据：
 
+## 指标白话释义
+
+- 主指标在衡量什么：
+- 次级指标在衡量什么：
+- 健康度/支持性指标该怎么读：
+
 ## 最终表现摘要
 
 - 主要结果摘要：
 - 最重要数字：
 - 报告边界：
 
+## 怎么看这些表
+
+- Table 1 负责回答什么：
+- Table 2 负责回答什么：
+- Table 3 负责回答什么：
+- Table 4 负责回答什么：
+
 ## Table 1
 
 - 作用：
@@ -1473,24 +1513,24 @@ ZH_CONTENT[path.join(".codex", "prompts", "lab-auto.md")] = codexPrompt(
 );
 
 ZH_CONTENT[path.join(".claude", "commands", "lab.md")] = claudeCommand(
-  "LAB",
+  "lab",
   "查看 /lab 研究工作流总览并选择合适阶段",
-  "workflow, research, overview",
-  "# `/lab` for Claude\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  调研 idea，定义问题与 failure case，归类 contribution 与 breakthrough level，对比现有方法，收束三个一眼就有意义的点，并在实现前保留 approval gate。\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- 先给简洁摘要，再决定是否写工件，最后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- 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 reviewer fidelity` 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 的 organizer enforcement。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：evaluator prompt registry、ingestion、parser。`\n"
+  "[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- 先给简洁摘要，再决定是否写工件，最后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- 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 reviewer fidelity` 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 的 organizer enforcement。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：evaluator prompt registry、ingestion、parser。`\n"
 );
 
-ZH_CONTENT[path.join(".claude", "commands", "lab", "data.md")] = claudeCommand(
-  "LAB: Data",
+ZH_CONTENT[path.join(".claude", "commands", "lab-data.md")] = claudeCommand(
+  "lab-data",
   "在进入规格前锁定数据集、下载来源和 benchmark 组合",
-  "workflow, research, data",
-  "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 `/lab:data`，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:data` 阶段。它必须把已批准的 idea 转成数据集与 benchmark 方案，记录每个候选数据集的年份、使用过它的论文、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。"
+  "dataset 或 benchmark target",
+  "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 `data` 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 `data` 阶段。它必须把已批准的 idea 转成数据集与 benchmark 方案，记录每个候选数据集的年份、使用过它的论文、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。"
 );
 
-ZH_CONTENT[path.join(".claude", "commands", "lab", "auto.md")] = claudeCommand(
-  "LAB: Auto",
+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",
-  "workflow, research, auto",
-  "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 `/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。"
+  "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。"
 );
 
 ZH_CONTENT[path.join(".codex", "skills", "lab", "SKILL.md")] = `---
@@ -2016,6 +2056,7 @@ ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "auto.md")] = `# \`/la
 - 先做输入归一化：把 \`Autonomy level L1/L2/L3\` 视为执行权限级别，把 \`Layer 3\`、\`Phase 1\`、\`Table 2\` 视为论文范围目标。
 - 如果用户同时提了论文层、实验 phase 和自治级别，先用一句话重述：objective、自治级别、terminal goal、scope、allowed modifications。
 - 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。
+- 当循环进入 \`report\` 时，要主动给出用户可读的白话总结，解释主指标、次级指标和主表作用；不要等用户额外发一句“解释这些指标”。
 - 当循环即将进入 \`write\`，且 \`paper_template_root\` 为空时：
   - 如果 \`paper_template_decision\` 是 \`unconfirmed\`，必须先追问一次：继续使用默认 scaffold，还是先接入模板目录
   - 如果用户选择默认 scaffold，就持久化 \`paper_template_decision: default-scaffold\`
@@ -2028,6 +2069,8 @@ ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "auto.md")] = `# \`/la
 
 ZH_CONTENT[path.join(".claude", "skills", "lab", "stages", "auto.md")] =
   ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "auto.md")];
+ZH_CONTENT[path.join(".claude", "skills", "lab", "stages", "report.md")] =
+  ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "report.md")];
 
 function getLocalizedContent(relativePath, lang) {
   if (lang !== "zh") {

package/lib/install.cjs

@@ -326,6 +326,8 @@ function detectLegacyPlatform(targetDir) {
     fs.existsSync(path.join(targetDir, ".codex", "prompts")) ||
     fs.existsSync(path.join(targetDir, ".codex", "skills", "lab"));
   const hasClaude =
+    fs.existsSync(path.join(targetDir, ".claude", "commands", "lab.md")) ||
+    fs.existsSync(path.join(targetDir, ".claude", "commands", "lab-idea.md")) ||
     fs.existsSync(path.join(targetDir, ".claude", "commands", "lab")) ||
     fs.existsSync(path.join(targetDir, ".claude", "skills", "lab"));
 
@@ -366,6 +368,7 @@ function detectLegacyLanguage(targetDir) {
 
   const probeFiles = [
     path.join(targetDir, ".codex", "prompts", "lab-idea.md"),
+    path.join(targetDir, ".claude", "commands", "lab-idea.md"),
     path.join(targetDir, ".claude", "commands", "lab", "idea.md"),
     path.join(targetDir, ".lab", ".managed", "templates", "idea.md"),
     path.join(targetDir, ".superlab", "templates", "idea.md"),
@@ -494,16 +497,16 @@ function localizeInstalledAssets(targetDir, lang, { newlyCreatedProjectOwnedPath
     path.join(".codex", "prompts", "lab-report.md"),
     path.join(".codex", "prompts", "lab-write.md"),
     path.join(".claude", "commands", "lab.md"),
-    path.join(".claude", "commands", "lab", "idea.md"),
-    path.join(".claude", "commands", "lab", "data.md"),
-    path.join(".claude", "commands", "lab", "auto.md"),
-    path.join(".claude", "commands", "lab", "framing.md"),
-    path.join(".claude", "commands", "lab", "spec.md"),
-    path.join(".claude", "commands", "lab", "run.md"),
-    path.join(".claude", "commands", "lab", "iterate.md"),
-    path.join(".claude", "commands", "lab", "review.md"),
-    path.join(".claude", "commands", "lab", "report.md"),
-    path.join(".claude", "commands", "lab", "write.md"),
+    path.join(".claude", "commands", "lab-idea.md"),
+    path.join(".claude", "commands", "lab-data.md"),
+    path.join(".claude", "commands", "lab-auto.md"),
+    path.join(".claude", "commands", "lab-framing.md"),
+    path.join(".claude", "commands", "lab-spec.md"),
+    path.join(".claude", "commands", "lab-run.md"),
+    path.join(".claude", "commands", "lab-iterate.md"),
+    path.join(".claude", "commands", "lab-review.md"),
+    path.join(".claude", "commands", "lab-report.md"),
+    path.join(".claude", "commands", "lab-write.md"),
     path.join(".codex", "skills", "lab", "SKILL.md"),
     path.join(".codex", "skills", "lab", "stages", "idea.md"),
     path.join(".codex", "skills", "lab", "stages", "data.md"),

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

@@ -0,0 +1,13 @@
+---
+name: "lab-auto"
+description: Orchestrate approved lab execution stages inside a bounded autonomous loop
+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`.
+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-data.md

@@ -0,0 +1,10 @@
+---
+name: "lab-data"
+description: Select datasets and benchmark packages with explicit source, year, and paper-usage audit
+argument-hint: dataset or benchmark selection target
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-data` 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 `data` stage of the lab workflow. It must turn the approved idea into an approved dataset package with dataset year, papers that used each dataset, source audit, download plan, explicit benchmark-role rationale for classic-public, recent-strong-public, and claim-specific benchmarks, and explicit comparison rationale for canonical baselines, strong historical baselines, recent strong public methods, and closest prior work.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-framing"
+description: Lock paper-facing naming, title, and contribution framing before drafting
+argument-hint: naming, title, or contribution framing target
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-framing` 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 `framing` stage of the lab workflow. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-idea"
+description: Research and refine a paper or experiment idea before specification
+argument-hint: idea or research problem
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-idea` 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 `idea` stage of the lab workflow. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-iterate"
+description: Run bounded Ralph-style experiment iterations with fixed success criteria
+argument-hint: iteration mission
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-iterate` 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 `iterate` stage of the lab workflow. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-report"
+description: Produce the final report from validated iteration artifacts
+argument-hint: report context
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-report` 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 `report` stage of the lab workflow. It must produce a user-facing final report plus the managed `main-tables.md` artifact, explicitly carry the approved primary and secondary metrics forward, explain the selected metrics in plain language, say which metrics are only health or support metrics, and explain what each main table proves or does not prove.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-review"
+description: Review a research plan or result in reviewer mode
+argument-hint: artifact or result to review
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-review` 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 `review` stage of the lab workflow. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-run"
+description: Execute the smallest meaningful experiment and normalize its output
+argument-hint: run context
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-run` 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 `run` stage of the lab workflow. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-spec"
+description: Convert an approved idea into a lab change directory
+argument-hint: approved idea context
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-spec` 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 `spec` stage of the lab workflow. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -0,0 +1,10 @@
+---
+name: "lab-write"
+description: Turn validated research artifacts into paper sections with small evidence-bound revisions
+argument-hint: section or writing target
+---
+
+Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
+
+Execute the requested `/lab-write` 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 `write` stage of the lab workflow. It requires an approved framing artifact from the `framing` stage, must read the matching section reference from `.claude/skills/lab/references/paper-writing/`, and for `abstract`, `introduction`, or `method` it must also read `.claude/skills/lab/references/paper-writing/examples/index.md` plus the matching examples index and 1-2 concrete example files. Then it should run `paper-review.md` and `does-my-writing-flow-source.md`, build a mini-outline, and revise only one section.

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

@@ -1,66 +1,70 @@
 ---
-name: "LAB"
-description: Overview of the /lab research workflow and stage selection
-category: Workflow
-tags: [workflow, research, overview]
+name: "lab"
+description: Overview and dispatcher for the lab research workflow in Claude Code
+argument-hint: [stage] [target]
 ---
 
 # `/lab` for Claude
 
-`/lab` is a strict research workflow command family. Use the same repository artifacts and stage boundaries every time.
+`/lab` is the Claude Code dispatcher for the strict lab research workflow. In Claude Code, use either:
 
-## Subcommands
+- `/lab <stage> ...` for the root dispatcher
+- `/lab-idea`, `/lab-data`, `/lab-auto`, `/lab-framing`, `/lab-spec`, `/lab-run`, `/lab-iterate`, `/lab-review`, `/lab-report`, `/lab-write` for direct stage aliases
 
-- `/lab:idea`
+Use the same repository artifacts and stage boundaries every time.
+
+## Stage Aliases
+
+- `/lab idea ...` or `/lab-idea`
   Research the idea, define the problem and failure case, classify the contribution and breakthrough level, compare against existing methods, end with three meaningful points, and keep an explicit approval gate before any implementation.
 
-- `/lab:data`
+- `/lab data ...` or `/lab-data`
   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`
+- `/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.
 
-- `/lab:framing`
+- `/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.
 
-- `/lab:spec`
+- `/lab spec ...` or `/lab-spec`
   Convert the approved idea into one lab change directory under `.lab/changes/<change-id>/`, then draft `proposal`, `design`, `spec`, and `tasks` inside that directory.
 
-- `/lab:run`
+- `/lab run ...` or `/lab-run`
   Execute the smallest useful validation run, register it, and produce the first normalized evaluation summary.
 
-- `/lab:iterate`
+- `/lab iterate ...` or `/lab-iterate`
   Run bounded Ralph Wiggum style experiment loops with a frozen mission, explicit thresholds, deterministic verification commands, `completion_promise`, and per-round reports.
 
-- `/lab:review`
+- `/lab review ...` or `/lab-review`
   Audit documents or results in reviewer mode. Start with a short summary, then output findings, fatal flaws, fix priority, and residual risks.
 
-- `/lab:report`
+- `/lab report ...` or `/lab-report`
   Generate the final research report from accumulated runs and iteration artifacts.
 
-- `/lab:write`
+- `/lab write ...` or `/lab-write`
   Turn stable report artifacts into paper sections through small, evidence-bound writing rounds using the vendored paper-writing references under the installed `lab` skill.
   On the first manuscript-writing round, if `paper_template_root` is empty, explicitly ask once whether to stay on the managed default LaTeX scaffold or attach a template directory first; persist the user's default-scaffold choice before continuing.
 
 ## Dispatch Rules
 
 - Always use `skills/lab/SKILL.md` as the workflow contract.
-- When the user explicitly invokes `/lab:<stage>`, execute that stage now against the provided argument instead of only recommending another `/lab` stage.
+- 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 summary, then decide whether to write artifacts, 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.
-- `/lab:spec` is not complete until the approved change is frozen under `.lab/changes/<change-id>/`.
-- `/lab:spec` should inherit the approved dataset package from `.lab/context/data-decisions.md`.
-- Never skip directly from `/lab:idea` to code.
-- `/lab:iterate` requires a normalized summary from `scripts/eval_report.py`.
-- `/lab:run`, `/lab:iterate`, `/lab:auto`, and `/lab:report` should all follow `.lab/context/eval-protocol.md`, including its recorded sources for metrics and comparison implementations.
-- `/lab:write` requires an approved framing artifact from `/lab:framing`.
-- `/lab:write` requires stable report artifacts, a mini-outline, the active section guide, `paper-review.md`, and `does-my-writing-flow-source.md`, and should only change one section per round.
+- `spec` is not complete until the approved change is frozen under `.lab/changes/<change-id>/`.
+- `spec` should inherit the approved dataset package from `.lab/context/data-decisions.md`.
+- Never skip directly from `idea` to code.
+- `iterate` requires a normalized summary from `scripts/eval_report.py`.
+- `run`, `iterate`, `auto`, and `report` should all follow `.lab/context/eval-protocol.md`, including its recorded sources for metrics and comparison implementations.
+- `write` requires an approved framing artifact from the `framing` stage.
+- `write` requires stable report artifacts, a mini-outline, the active section guide, `paper-review.md`, and `does-my-writing-flow-source.md`, and should only change one section per round.
 
-## How to Ask for `/lab:auto`
+## How to Ask for `/lab auto`
 
 - 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 reviewer fidelity` should not be interpreted as `Autonomy level L3`.
-- A good `/lab:auto` request should name:
+- A good `/lab auto` request should name:
   - the objective
   - the autonomy level
   - the terminal goal
@@ -68,4 +72,4 @@ tags: [workflow, research, overview]
   - the allowed modifications
 - If the repository workflow language is Chinese, summaries, checklist items, task labels, and progress updates should be written in Chinese unless a code identifier or file path must stay literal.
 - Good example:
-  - `/lab:auto Autonomy level L2. Objective: advance paper layer 3 organizer enforcement. Terminal goal: task-completion. Scope: bounded protocol, tests, minimal implementation, and one small run. Allowed modifications: evaluator prompt registry, ingestion, and parser only.`
+  - `/lab auto Autonomy level L2. Objective: advance paper layer 3 organizer enforcement. Terminal goal: task-completion. Scope: bounded protocol, tests, minimal implementation, and one small run. Allowed modifications: evaluator prompt registry, ingestion, and parser only.`

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

@@ -6,4 +6,4 @@ argument-hint: report context
 Use the installed `lab` skill at `.codex/skills/lab/SKILL.md`.
 
 Execute the requested `/lab:report` 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:report` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.
+This command runs the `/lab:report` stage. It must produce a user-facing final report plus the managed `main-tables.md` artifact, explicitly carry the approved primary and secondary metrics forward, explain the selected metrics in plain language, say which metrics are only health or support metrics, and explain what each main table proves or does not prove.

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

@@ -1,8 +1,11 @@
 # Final Report
 
-## Overview
+## Reader Summary
 
-Summarize the method and overall outcome.
+- One-sentence conclusion:
+- What is validated:
+- What is still unproven:
+- Biggest reporting risk:
 
 ## Selected Metrics
 
@@ -10,6 +13,12 @@ Summarize the method and overall outcome.
 - Secondary metrics:
 - Required terminal evidence:
 
+## Metric Guide
+
+- Primary metric plain-language explanation:
+- Secondary metric plain-language explanation:
+- Health or support metrics and why they are not the main claim:
+
 ## Experiment Setup
 
 - Datasets:
@@ -17,6 +26,13 @@ Summarize the method and overall outcome.
 - Baselines:
 - Metrics:
 
+## How to Read the Main Tables
+
+- Table 1 is for:
+- Table 2 is for:
+- Table 3 is for:
+- Table 4 is for:
+
 ## Main Tables
 
 - Managed main tables artifact: `<deliverables_root>/main-tables.md`

package/package-assets/shared/lab/.managed/templates/main-tables.md

@@ -1,17 +1,36 @@
 # Main Tables
 
+## Reader Summary
+
+- User-facing takeaway:
+- What the tables prove:
+- What the tables do not yet prove:
+
 ## Selected Metrics
 
 - Primary metrics:
 - Secondary metrics:
 - Required terminal evidence:
 
+## Metric Guide
+
+- Primary metric plain-language explanation:
+- Secondary metric plain-language explanation:
+- Health or support metrics and how to read them:
+
 ## Final Performance Summary
 
 - Main result summary:
 - Most important numbers:
 - Reporting caveat:
 
+## How to Read These Tables
+
+- Table 1 is for:
+- Table 2 is for:
+- Table 3 is for:
+- Table 4 is for:
+
 ## Table 1
 
 - Purpose:

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

@@ -108,6 +108,7 @@
 - Then ask at most one clarifying question if a blocking field is still missing.
 - If `.lab/config/workflow.json` sets the workflow language to Chinese, write summaries, options, checklist items, task labels, and progress updates in Chinese unless a file path, code identifier, or literal metric name must remain unchanged.
 - When the loop reaches `report`, apply the same workflow-language rule to `report.md` and the managed `main-tables.md` artifact.
+- When the loop reaches `report`, proactively deliver a user-facing plain-language summary of the selected metrics, what they mean, what the tables prove, and what remains unproven. Do not wait for a separate user request asking for interpretation.
 - When the loop is about to enter `write` and `paper_template_root` is empty:
   - if `paper_template_decision` is `unconfirmed`, ask one explicit question: continue with the default scaffold or attach a template directory first
   - if the user chooses the default scaffold, persist `paper_template_decision: default-scaffold`

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

@@ -2,11 +2,14 @@
 
 ## Required Output
 
+- reader summary for the user
 - method overview
 - selected metrics summary
+- plain-language metric guide
 - experiment setup
 - validated main results
 - managed main tables artifact under `<deliverables_root>/main-tables.md`
+- how-to-read-the-tables guide
 - ablations
 - failed attempts
 - limitations
@@ -34,6 +37,8 @@
 - Structure tables, gates, and main claims against the approved evaluation protocol.
 - Do not restate metric definitions, baseline behavior, or comparison implementations from memory; use the approved evaluation protocol and its recorded sources.
 - Carry the approved `Primary metrics`, `Secondary metrics`, and `Required terminal evidence` into both the report and the managed main-tables artifact.
+- Explain the selected primary and secondary metrics in plain language for the user: what each metric measures, whether higher or lower is better, and whether it is a main result metric or only a health/support metric.
+- If coverage, completeness, confidence, or similar health metrics appear, explicitly say that they describe experimental reliability rather than the main scientific effect.
 - If the report depends on a deviation from an original metric or implementation, state that deviation explicitly instead of smoothing it over.
 - 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.
@@ -42,6 +47,8 @@
 ## Interaction Contract
 
 - Start with a concise summary of the campaign outcome, the selected primary and secondary metrics, the strongest supported claim, and the biggest reporting risk.
+- Proactively deliver a user-readable plain-language summary when the stage is reached from `/lab:auto`; do not wait for a separate follow-up request asking what the metrics or tables mean.
+- Treat `report.md` as a user-facing artifact rather than an internal dump. Prefer plain-language explanations before jargon, and explain each metric the first time it matters.
 - If a missing assumption would change report interpretation, ask one clarifying question at a time.
 - If there are multiple defensible report framings, present 2-3 approaches with trade-offs and recommend the most evidence-faithful framing before writing.
 - Keep an approval gate when the reporting frame would materially affect what the paper later claims.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "superlab",
-  "version": "0.1.15",
+  "version": "0.1.17",
   "description": "Strict /lab research workflow installer for Codex and Claude",
   "keywords": [
     "codex",

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

@@ -1,14 +0,0 @@
----
-name: "LAB: Auto"
-description: Orchestrate approved lab execution stages inside a bounded autonomous loop
-category: Workflow
-tags: [workflow, research, auto]
----
-
-Use the installed `lab` skill at `.claude/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`.
-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/data.md

@@ -1,11 +0,0 @@
----
-name: "LAB: Data"
-description: Select datasets and benchmark packages with explicit source, year, and paper-usage audit
-category: Workflow
-tags: [workflow, research, data]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:data` 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:data` stage. It must turn the approved idea into an approved dataset package with dataset year, papers that used each dataset, source audit, download plan, explicit benchmark-role rationale for classic-public, recent-strong-public, and claim-specific benchmarks, and explicit comparison rationale for canonical baselines, strong historical baselines, recent strong public methods, and closest prior work.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Framing"
-description: Lock paper-facing naming, title, and contribution framing before drafting
-category: Workflow
-tags: [workflow, research, framing]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:framing` 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:framing` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Idea"
-description: Research and refine a paper or experiment idea before specification
-category: Workflow
-tags: [workflow, research, idea]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:idea` 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:idea` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Iterate"
-description: Run bounded Ralph-style experiment iterations with fixed success criteria
-category: Workflow
-tags: [workflow, research, iterate]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:iterate` 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:iterate` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Report"
-description: Produce the final report from validated iteration artifacts
-category: Workflow
-tags: [workflow, research, report]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:report` 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:report` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Review"
-description: Review a research plan or result in reviewer mode
-category: Workflow
-tags: [workflow, research, review]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:review` 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:review` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Run"
-description: Execute the smallest meaningful experiment and normalize its output
-category: Workflow
-tags: [workflow, research, run]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:run` 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:run` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Spec"
-description: Convert an approved idea into a lab change directory
-category: Workflow
-tags: [workflow, research, spec]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:spec` 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:spec` stage. Follow the installed skill, stage guide, and the project assets under `.lab/`.

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

@@ -1,11 +0,0 @@
----
-name: "LAB: Write"
-description: Turn validated research artifacts into paper sections with small evidence-bound revisions
-category: Workflow
-tags: [workflow, research, writing]
----
-
-Use the installed `lab` skill at `.claude/skills/lab/SKILL.md`.
-
-Execute the requested `/lab:write` 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:write` stage. It requires an approved framing artifact from `/lab:framing`, must read the matching section reference from `.claude/skills/lab/references/paper-writing/`, and for `abstract`, `introduction`, or `method` it must also read `.claude/skills/lab/references/paper-writing/examples/index.md` plus the matching examples index and 1-2 concrete example files. Then it should run `paper-review.md` and `does-my-writing-flow-source.md`, build a mini-outline, and revise only one section.