superlab 0.1.62 → 0.1.63

package/lib/i18n.cjs

@@ -2092,8 +2092,10 @@ ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "write.md")] = `# \`/l
 - 最终稿必须是 LaTeX。
 - 普通起草轮次先跟随 \`workflow_language\`。
 - 当 \`workflow_language\` 与 \`paper_language\` 不一致时，把 workflow-language 论文层当成普通写作轮次的默认工作层。
+- 先根据 \`.lab/config/workflow.json\` 解析现役 paper topology：canonical 根目录是 \`<deliverables_root>/paper/\`，当 workflow-language 现役时，它的根目录是 \`<deliverables_root>/paper/workflow-language/\`。
 - 普通写作轮次仍然只改一个目标层，不要静默同时刷新两种语言层。
 - 如果用户明确点名了某个文件或层级，就把它当成当前轮次唯一目标，除非用户另外明确要求同步。
+- 先给目标路径做角色分类。只有现役层路径才算受管 manuscript round；像 \`review_zh\`、\`translation_zh\`、\`sections_zh\` 或旧的 \`deliverables/.../workflow-language/*.md\` 这类路径，只能当作 out-of-band / legacy 编辑，不能再冒充现役论文层。
 - 如果 workflow-language 论文层现役而本轮仍然写 canonical manuscript，必须在 write iteration artifact 里说明为什么这次 canonical-only 写作是合理的。
 - 如果 \`paper_language_finalization_decision=convert-to-paper-language\`，显式的 canonical manuscript 工作可以直接落在 \`paper_language\`，但这并不意味着只要 workflow-language 仍然现役，canonical 就变成普通轮次的默认工作层。
 - 把 workflow-language 论文层当成正式持久化产物，而不是 review 层。
@@ -2979,7 +2981,9 @@ ZH_CONTENT[path.join(".codex", "skills", "lab", "stages", "auto.md")] = `# \`/la
 - 普通 \`write\` 起草轮次先跟随 \`workflow_language\`。
 - 普通 \`write\` 起草轮次里的 manuscript 仍然是 LaTeX，所以普通 manuscript \`.tex\` sections 也必须先保留在 \`workflow_language\`，还不是 \`paper_language\` 版本。
 - 当 \`workflow_language\` 与 \`paper_language\` 不一致时，把 workflow-language 论文层当成普通写作轮次的默认工作层。
+- 先根据 \`.lab/config/workflow.json\` 解析现役 paper topology：canonical 根目录是 \`<deliverables_root>/paper/\`，当 workflow-language 现役时，它的根目录是 \`<deliverables_root>/paper/workflow-language/\`。
 - 普通写作轮次仍然只改一个目标层，不要在没有明确要求的情况下同时刷新 canonical manuscript 和 workflow-language 论文层。
+- 先给目标路径做角色分类。只有现役层路径才算受管 manuscript round；像 \`review_zh\`、\`translation_zh\`、\`sections_zh\` 或旧的 \`deliverables/.../workflow-language/*.md\` 这类路径，只能当作 out-of-band / legacy 编辑，不能再冒充现役论文层。
 - 把 workflow-language 论文层当成正式持久化产物，而不是 review 层。
 - 如果 workflow-language 论文层现役而本轮仍然写 canonical manuscript，必须在最新 write iteration 里说明为什么这次 canonical-only 写作是合理的。
 - 如果当前写作目标是最终导出、\`workflow_language\` 与 \`paper_language\` 不一致，且 \`paper_language_finalization_decision\` 还是 \`unconfirmed\`，就在最终定稿前先完成并保留 workflow-language 论文层，再追问一次：保持当前 workflow language，还是转换成 \`paper_language\`

package/lib/lab_write_contract.json

@@ -1,6 +1,6 @@
 {
   "stage_prompt": {
-    "codex_en": "This command runs the `/lab:write` stage. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\\resizebox{\\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\\tabcolsep` conservatively; only use `\\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\\scriptsize` or `\\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.",
+    "codex_en": "This command runs the `/lab:write` stage. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Resolve the active paper topology from `.lab/config/workflow.json` before drafting: the active canonical root is `<deliverables_root>/paper/`, and when workflow-language is active its root is `<deliverables_root>/paper/workflow-language/`. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. Classify the named target path before editing it. Only active-layer targets count as managed manuscript rounds; legacy side layers such as `review_zh`, `translation_zh`, `sections_zh`, or stale `deliverables/.../workflow-language/*.md` paths are out-of-band/legacy edits and must not silently replace the active paper topology. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\\resizebox{\\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\\tabcolsep` conservatively; only use `\\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\\scriptsize` or `\\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, the active canonical/workflow-language roots, the resolved target path role, any out-of-band justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.",
     "claude_en": "This command runs the `write` stage of the lab workflow. Use `.claude/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\\resizebox{\\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\\tabcolsep` conservatively; only use `\\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\\scriptsize` or `\\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.",
     "codex_zh": "本命令运行 `/lab:write` 阶段。把 `.codex/skills/lab/stages/write.md` 当成模板选择、paper-plan、section 参考、校验 gate、资产覆盖和最终 manuscript 规则的单一来源。读取与当前 section 对应的 paper-writing reference、`section-style-policies.md` 里与当前 section 对应的风格块，以及 bundled example-bank 文件，一次只修改一个 section；普通草稿轮次把写作校验当 warning，最终定稿或导出轮次必须满足 write-stage 的接受 gate。普通起草轮次先跟随 `workflow_language`，普通 `.tex` section 草稿也必须先停留在 `workflow_language`，不能把 `paper_language` 当成默认草稿语言，并把 workflow-language 论文层当成正式持久化产物，而不是 review 层。当 `workflow_language` 与 `paper_language` 不一致时，把 workflow-language 论文层当成普通写作轮次的默认工作层。普通写作轮次仍然只改一个目标层，不要静默同时刷新两种语言层。如果用户明确点名了某个文件或层级，就把它当成当前轮次唯一目标，除非用户另外明确要求同步。如果 workflow-language 论文层现役而本轮仍然写 canonical manuscript，必须在 write iteration artifact 里说明为什么这次 canonical-only 写作是合理的。如果 `paper_language_finalization_decision=convert-to-paper-language`，显式的 canonical manuscript 工作可以直接落在 `paper_language`，但这并不意味着只要 workflow-language 仍然现役，canonical 就变成普通轮次的默认工作层。workflow-language 层必须是一套完整的 LaTeX 镜像，至少包含 `workflow-language/main.tex`、`workflow-language/references.bib`、`workflow-language/sections/*.tex`、`workflow-language/tables/*.tex`、`workflow-language/figures/*.tex` 和 `workflow-language/analysis/analysis-asset.tex`。不要再把新的 workflow-language 输出写到已经废弃的 review 层路径，例如 `docs/lab/paper/review_zh/`。把 `.lab/writing/terminology-glossary.md` 作为写作期 glossary，用来沉淀全称、批准缩写、对外解释和可接受别名。无论当前语言是什么，都要满足同一套学术可读性标准：如果本轮引入或改写了关键术语、缩写、指标名、机制名或系统标签，就先写全称；如果后面要复用短写，就在首次出现时定义；同时说明它是什么、为什么在这里重要，保持一个概念只有一个 paper-facing 名称，并尽量避免新造的连字符拼接标签。相同的首次解释规则也适用于表头、caption、表注和图注；如果术语第一次出现在表里，就必须在同一张表里局部解释。不要假定当前 section 可以无限继续压句子；在同一个 section 上做新的 tighten、compress 或 polish 之前，先过一遍 section-level acceptance gate。这个 gate 至少要显式确认命名一致性、前后文一致性、claim / metric / ranking 与当前证据的一致性、局部清晰度、局部简洁度，以及 section-style compliance。当前 section 的写作表达必须先遵守 `section-style-policies.md` 里的鼓励表达、谨慎表达和禁用表达；只要还有禁用表达或禁用写法残留，本轮就不能算通过 section-level acceptance gate。如果本轮改动的是论文当前的 canonical 实验或评测协议（例如 split 比例、训练/测试规模、seed 或 split 数量、benchmark 集合、主表评测口径），默认把它当成主协议替换，而不是补充实验；除非用户明确说这是 supplementary 或 appendix-only。遇到这种协议替换时，先做全文影响审计（paper-wide impact audit），列出哪些 section 和资产已经过期，优先更新影响最大的 canonical section / asset，再继续润色；不要默认去做 translation/workflow-layer 同步，除非用户明确要求，或者语言最终定稿流程要求这样做。只有在用户明确要求跨语言同步，或者 final-draft/export 的语言最终定稿步骤确实要求两层一起刷新时，才允许同一轮同时修改 canonical manuscript 和 workflow-language 论文层。不要把普通 tighten/compress/polish 请求理解成“顺手同步 companion”。如果当前是导出或远程发布轮次，且 `paper_language_finalization_decision=convert-to-paper-language`，默认把 workflow-language 论文层一起带进导出的或推送的 bundle。只有在用户明确要求只导出 canonical，或者远程目标明确不允许附带额外文件时，才允许 canonical-only export。如果任何一项还没过，就先解决这个 blocker，不要默认继续 prose polish，也不要把下一步建议自动写成“再收紧一轮”。主表必须局部自足：读者只看表题、表头、表注和邻近引入或解读，就应当知道每行是什么、每列是什么、指标方向是什么，以及需要的单位、分母或触发条件。表头可以短，但表里的缩写必须在同表局部展开。如果 Method 或 Experiments 承诺了一组指标，主表就必须直接展示这些指标，或者明确标成 appendix-only 并解释原因；如果某个指标因为恒为零、冗余或只放附录而不单列，也必须在表注里显式说明，不能静默省略。不要把 `\\resizebox{\\linewidth}{!}{...}` 当成主表默认的宽度解决方案。主表如果超宽，先重构表：缩短表头但保留局部解释，把次要指标移到 appendix-only，减少或拆分列，最后才保守调整 `\\tabcolsep`；只有这些都做过仍然超宽时，才允许把 `\\resizebox` 当成最后手段，而且必须在同一张表的表注里解释宽度处理理由。不要把 `\\scriptsize` 或 `\\tiny` 当成主表默认的适配策略。内部标识符默认不要进入 reader-facing prose；若必须出现，只能在完成一次读者映射后使用，并把 terminology-clarity 自检、section-level acceptance gate、section-style policy 合规检查、protocol/scope impact audit、export 或 remote bundle 审计、当前轮次目标层、workflow-language 现役时的 canonical-only 理由、任何跨语言同步理由，以及 table-semantics 审计一起写进 write iteration artifact。如果当前稿件将从托管默认 scaffold 开始，且还没有模板决定，就先追问一次：继续使用默认 scaffold，还是先接入模板目录。如果进入最终定稿时 `workflow_language` 与 `paper_language` 不一致，就先完成并保留 workflow-language 论文层，再追问一次：保持当前语言，还是把 canonical manuscript 转成 `paper_language`；先持久化这个决定，再在最新 write iteration 里记录语言决策和 workflow-language 论文层路径，最后才允许按该语言修改最终稿。",
     "claude_zh": "本命令运行 lab workflow 的 `write` 阶段。把 `.claude/skills/lab/stages/write.md` 当成模板选择、paper-plan、section 参考、校验 gate、资产覆盖和最终 manuscript 规则的单一来源。读取与当前 section 对应的 paper-writing reference、`section-style-policies.md` 里与当前 section 对应的风格块，以及 bundled example-bank 文件，一次只修改一个 section；普通草稿轮次把写作校验当 warning，最终定稿或导出轮次必须满足 write-stage 的接受 gate。普通起草轮次先跟随 `workflow_language`，普通 `.tex` section 草稿也必须先停留在 `workflow_language`，不能把 `paper_language` 当成默认草稿语言，并把 workflow-language 论文层当成正式持久化产物，而不是 review 层。当 `workflow_language` 与 `paper_language` 不一致时，把 workflow-language 论文层当成普通写作轮次的默认工作层。普通写作轮次仍然只改一个目标层，不要静默同时刷新两种语言层。如果用户明确点名了某个文件或层级，就把它当成当前轮次唯一目标，除非用户另外明确要求同步。如果 workflow-language 论文层现役而本轮仍然写 canonical manuscript，必须在 write iteration artifact 里说明为什么这次 canonical-only 写作是合理的。如果 `paper_language_finalization_decision=convert-to-paper-language`，显式的 canonical manuscript 工作可以直接落在 `paper_language`，但这并不意味着只要 workflow-language 仍然现役，canonical 就变成普通轮次的默认工作层。workflow-language 层必须是一套完整的 LaTeX 镜像，至少包含 `workflow-language/main.tex`、`workflow-language/references.bib`、`workflow-language/sections/*.tex`、`workflow-language/tables/*.tex`、`workflow-language/figures/*.tex` 和 `workflow-language/analysis/analysis-asset.tex`。不要再把新的 workflow-language 输出写到已经废弃的 review 层路径，例如 `docs/lab/paper/review_zh/`。把 `.lab/writing/terminology-glossary.md` 作为写作期 glossary，用来沉淀全称、批准缩写、对外解释和可接受别名。无论当前语言是什么，都要满足同一套学术可读性标准：如果本轮引入或改写了关键术语、缩写、指标名、机制名或系统标签，就先写全称；如果后面要复用短写，就在首次出现时定义；同时说明它是什么、为什么在这里重要，保持一个概念只有一个 paper-facing 名称，并尽量避免新造的连字符拼接标签。相同的首次解释规则也适用于表头、caption、表注和图注；如果术语第一次出现在表里，就必须在同一张表里局部解释。不要假定当前 section 可以无限继续压句子；在同一个 section 上做新的 tighten、compress 或 polish 之前，先过一遍 section-level acceptance gate。这个 gate 至少要显式确认命名一致性、前后文一致性、claim / metric / ranking 与当前证据的一致性、局部清晰度、局部简洁度，以及 section-style compliance。当前 section 的写作表达必须先遵守 `section-style-policies.md` 里的鼓励表达、谨慎表达和禁用表达；只要还有禁用表达或禁用写法残留，本轮就不能算通过 section-level acceptance gate。如果本轮改动的是论文当前的 canonical 实验或评测协议（例如 split 比例、训练/测试规模、seed 或 split 数量、benchmark 集合、主表评测口径），默认把它当成主协议替换，而不是补充实验；除非用户明确说这是 supplementary 或 appendix-only。遇到这种协议替换时，先做全文影响审计（paper-wide impact audit），列出哪些 section 和资产已经过期，优先更新影响最大的 canonical section / asset，再继续润色；不要默认去做 translation/workflow-layer 同步，除非用户明确要求，或者语言最终定稿流程要求这样做。只有在用户明确要求跨语言同步，或者 final-draft/export 的语言最终定稿步骤确实要求两层一起刷新时，才允许同一轮同时修改 canonical manuscript 和 workflow-language 论文层。不要把普通 tighten/compress/polish 请求理解成“顺手同步 companion”。如果当前是导出或远程发布轮次，且 `paper_language_finalization_decision=convert-to-paper-language`，默认把 workflow-language 论文层一起带进导出的或推送的 bundle。只有在用户明确要求只导出 canonical，或者远程目标明确不允许附带额外文件时，才允许 canonical-only export。如果任何一项还没过，就先解决这个 blocker，不要默认继续 prose polish，也不要把下一步建议自动写成“再收紧一轮”。主表必须局部自足：读者只看表题、表头、表注和邻近引入或解读，就应当知道每行是什么、每列是什么、指标方向是什么，以及需要的单位、分母或触发条件。表头可以短，但表里的缩写必须在同表局部展开。如果 Method 或 Experiments 承诺了一组指标，主表就必须直接展示这些指标，或者明确标成 appendix-only 并解释原因；如果某个指标因为恒为零、冗余或只放附录而不单列，也必须在表注里显式说明，不能静默省略。不要把 `\\resizebox{\\linewidth}{!}{...}` 当成主表默认的宽度解决方案。主表如果超宽，先重构表：缩短表头但保留局部解释，把次要指标移到 appendix-only，减少或拆分列，最后才保守调整 `\\tabcolsep`；只有这些都做过仍然超宽时，才允许把 `\\resizebox` 当成最后手段，而且必须在同一张表的表注里解释宽度处理理由。不要把 `\\scriptsize` 或 `\\tiny` 当成主表默认的适配策略。内部标识符默认不要进入 reader-facing prose；若必须出现，只能在完成一次读者映射后使用，并把 terminology-clarity 自检、section-level acceptance gate、section-style policy 合规检查、protocol/scope impact audit、export 或 remote bundle 审计、当前轮次目标层、workflow-language 现役时的 canonical-only 理由、任何跨语言同步理由，以及 table-semantics 审计一起写进 write iteration artifact。如果当前稿件将从托管默认 scaffold 开始，且还没有模板决定，就先追问一次：继续使用默认 scaffold，还是先接入模板目录。如果进入最终定稿时 `workflow_language` 与 `paper_language` 不一致，就先完成并保留 workflow-language 论文层，再追问一次：保持当前语言，还是把 canonical manuscript 转成 `paper_language`；先持久化这个决定，再在最新 write iteration 里记录语言决策和 workflow-language 论文层路径，最后才允许按该语言修改最终稿。"

package/lib/paper_topology.cjs

@@ -0,0 +1,91 @@
+const fs = require("node:fs");
+const path = require("node:path");
+
+function readWorkflowConfig(targetDir) {
+  const configPath = path.join(targetDir, ".lab", "config", "workflow.json");
+  if (!fs.existsSync(configPath)) {
+    return {};
+  }
+  try {
+    return JSON.parse(fs.readFileSync(configPath, "utf8"));
+  } catch {
+    return {};
+  }
+}
+
+function resolvePaperTopology({ targetDir, config } = {}) {
+  const workflowConfig = config || readWorkflowConfig(targetDir);
+  const deliverablesRoot = path.resolve(
+    targetDir,
+    workflowConfig.deliverables_root || path.join("docs", "research")
+  );
+  const canonicalRoot = path.join(deliverablesRoot, "paper");
+  const workflowLanguageActive =
+    !!workflowConfig.workflow_language &&
+    !!workflowConfig.paper_language &&
+    workflowConfig.workflow_language !== workflowConfig.paper_language;
+  const workflowLanguageRoot = workflowLanguageActive
+    ? path.join(canonicalRoot, "workflow-language")
+    : "";
+
+  return {
+    targetDir,
+    workflowConfig,
+    deliverablesRoot,
+    canonicalRoot,
+    workflowLanguageRoot,
+    workflowLanguageActive,
+  };
+}
+
+function normalizeInputPath(filePath, targetDir) {
+  if (!filePath) {
+    return "";
+  }
+  return path.isAbsolute(filePath) ? path.normalize(filePath) : path.resolve(targetDir || process.cwd(), filePath);
+}
+
+function isWithin(candidate, root) {
+  if (!candidate || !root) {
+    return false;
+  }
+  const relative = path.relative(root, candidate);
+  return relative === "" || (!relative.startsWith("..") && !path.isAbsolute(relative));
+}
+
+function looksLikeLegacyPaperLayer(normalizedPath, topology) {
+  const parts = normalizedPath.split(path.sep).filter(Boolean);
+  if (parts.some((part) => /^review_/i.test(part) || /^translation_/i.test(part) || /^sections_/i.test(part))) {
+    return true;
+  }
+  if (parts.includes("deliverables") && (parts.includes("sections") || parts.includes("workflow-language"))) {
+    return !isWithin(normalizedPath, topology.canonicalRoot);
+  }
+  if (parts.includes("docs") && parts.includes("lab") && parts.includes("paper")) {
+    return !isWithin(normalizedPath, topology.canonicalRoot);
+  }
+  return false;
+}
+
+function classifyPaperPath(filePath, topologyInput) {
+  const topology = topologyInput && topologyInput.canonicalRoot
+    ? topologyInput
+    : resolvePaperTopology(topologyInput || {});
+  const normalizedPath = normalizeInputPath(filePath, topology.targetDir);
+
+  if (topology.workflowLanguageActive && isWithin(normalizedPath, topology.workflowLanguageRoot)) {
+    return { role: "active-workflow-language", normalizedPath };
+  }
+  if (isWithin(normalizedPath, topology.canonicalRoot)) {
+    return { role: "active-canonical", normalizedPath };
+  }
+  if (looksLikeLegacyPaperLayer(normalizedPath, topology)) {
+    return { role: "legacy-paper-layer", normalizedPath };
+  }
+  return { role: "out-of-band", normalizedPath };
+}
+
+module.exports = {
+  classifyPaperPath,
+  resolvePaperTopology,
+};

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

@@ -6,4 +6,4 @@ argument-hint: section or writing target
 Use the installed `lab` skill at `.codex/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. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.
+This command runs the `/lab:write` stage. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Resolve the active paper topology from `.lab/config/workflow.json` before drafting: the active canonical root is `<deliverables_root>/paper/`, and when workflow-language is active its root is `<deliverables_root>/paper/workflow-language/`. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. Classify the named target path before editing it. Only active-layer targets count as managed manuscript rounds; legacy side layers such as `review_zh`, `translation_zh`, `sections_zh`, or stale `deliverables/.../workflow-language/*.md` paths are out-of-band/legacy edits and must not silently replace the active paper topology. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, the active canonical/workflow-language roots, the resolved target path role, any out-of-band justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.

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

@@ -6,4 +6,4 @@ argument-hint: section or writing target
 Use the installed `lab` skill at `.codex/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. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.
+This command runs the `/lab:write` stage. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Resolve the active paper topology from `.lab/config/workflow.json` before drafting: the active canonical root is `<deliverables_root>/paper/`, and when workflow-language is active its root is `<deliverables_root>/paper/workflow-language/`. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. Classify the named target path before editing it. Only active-layer targets count as managed manuscript rounds; legacy side layers such as `review_zh`, `translation_zh`, `sections_zh`, or stale `deliverables/.../workflow-language/*.md` paths are out-of-band/legacy edits and must not silently replace the active paper topology. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, the active canonical/workflow-language roots, the resolved target path role, any out-of-band justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.

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

@@ -6,4 +6,4 @@ argument-hint: section or writing target
 Use the installed `lab` skill at `.codex/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. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.
+This command runs the `/lab:write` stage. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Resolve the active paper topology from `.lab/config/workflow.json` before drafting: the active canonical root is `<deliverables_root>/paper/`, and when workflow-language is active its root is `<deliverables_root>/paper/workflow-language/`. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. Classify the named target path before editing it. Only active-layer targets count as managed manuscript rounds; legacy side layers such as `review_zh`, `translation_zh`, `sections_zh`, or stale `deliverables/.../workflow-language/*.md` paths are out-of-band/legacy edits and must not silently replace the active paper topology. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, the active canonical/workflow-language roots, the resolved target path role, any out-of-band justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.

package/package-assets/codex/prompts/lab/357/274/232write.md

@@ -6,4 +6,4 @@ argument-hint: section or writing target
 Use the installed `lab` skill at `.codex/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. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.
+This command runs the `/lab:write` stage. Use `.codex/skills/lab/stages/write.md` as the single source of truth for template choice, paper-plan requirements, section references, validator gates, asset coverage, and final manuscript rules. Read the matching paper-writing reference, the current section block in `section-style-policies.md`, and any bundled example-bank files for the requested section, revise only one section, and keep draft rounds warning-only while final-draft or export rounds must satisfy the write-stage acceptance gates. Draft ordinary manuscript rounds in `workflow_language`, and ordinary `.tex` section drafts must stay in `workflow_language` instead of treating `paper_language` as the default draft language. When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as the default ordinary working layer. Resolve the active paper topology from `.lab/config/workflow.json` before drafting: the active canonical root is `<deliverables_root>/paper/`, and when workflow-language is active its root is `<deliverables_root>/paper/workflow-language/`. Ordinary write rounds should still edit one target paper layer at a time rather than silently refreshing both language layers. If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization. Classify the named target path before editing it. Only active-layer targets count as managed manuscript rounds; legacy side layers such as `review_zh`, `translation_zh`, `sections_zh`, or stale `deliverables/.../workflow-language/*.md` paths are out-of-band/legacy edits and must not silently replace the active paper topology. If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write iteration artifact. If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active. Treat the workflow-language paper layer as a real persisted artifact rather than a review layer, and preserve it as a full LaTeX mirror with `workflow-language/main.tex`, `workflow-language/references.bib`, `workflow-language/sections/*.tex`, `workflow-language/tables/*.tex`, `workflow-language/figures/*.tex`, and `workflow-language/analysis/analysis-asset.tex`. Do not write new workflow-language output to deprecated review-layer paths such as `docs/lab/paper/review_zh/`. Maintain `.lab/writing/terminology-glossary.md` as the write-stage glossary for full forms, approved short forms, reader-facing explanations, and aliases. Apply the same academic readability standard in every language: when the round introduces or revises key terms, abbreviations, metrics, mechanism names, or system labels, use the full form first, define any short form at first mention, explain what the term is and why it matters here, keep one natural-language paper-facing name per concept, use natural-language full names in prose, do not use labels containing `_` or `-` in reader-facing prose, apply the same first-mention rule to table headers, table captions, table notes, and figure captions or labels, do not assume a fixed drafting order such as Method before Experiments, add a local naming bridge when a section uses canonical short names before their defining section has been drafted, and reuse the canonical label instead of replacing it with a narrative alias. Follow the current section's encouraged, discouraged, and banned expression lists from `section-style-policies.md`; section-specific banned expressions take priority over prose-polish goals. Before any additional tighten, compress, or polish pass on the same section, run a section-level acceptance gate first. That gate must explicitly confirm naming consistency, adjacent-section consistency, claim, metric, and ranking consistency with the current evidence, local clarity, local concision, and section-style compliance. If the round changes the paper's canonical experiment or evaluation protocol, treat that change as a canonical replacement unless the user explicitly scoped it as supplementary or appendix-only, run a paper-wide impact audit before more polishing, update the highest-impact stale sections and assets first, and do not default to translation/workflow-layer sync work unless the user explicitly asked for it or the language-finalization workflow requires it. Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together. Do not treat a routine tighten/compress/polish request as an instruction to sync the workflow-language companion. For export or remote-publication rounds, if `paper_language_finalization_decision=convert-to-paper-language`, include the workflow-language paper layer in the exported or pushed bundle by default. Allow canonical-only export or remote publication only when the user explicitly asked for it or when the remote target forbids extra files. If any gate item is unresolved, or if a banned expression or move from the current section policy remains, spend the round fixing that blocker instead of polishing sentences further, and do not default the next-step recommendation to another polish pass. Main tables must be locally self-contained: the title, header, note, and adjacent prose should tell the reader what each row and column means, the metric direction, and any relevant unit, denominator, or event condition. Short headers remain allowed, but abbreviations in paper-facing tables must be expanded locally in the same table. If Method or Experiments prose promises a metric family, the main table set must either expose those metrics directly or explicitly mark the missing ones as appendix-only and explain why. If a metric is measured but omitted because it is zero everywhere, redundant, or appendix-only, state that disposition explicitly in the table note instead of silently dropping it. Do not treat `\resizebox{\linewidth}{!}{...}` as the default way to fit a main table. Fit main tables by redesign first: shorten headers, move secondary metrics out of the main table, reduce or split columns, then adjust `\tabcolsep` conservatively; only use `\resizebox` as a last resort, keep width changes readable, and explain the width-control rationale locally in the same table note. Do not use `\scriptsize` or `\tiny` as the default main-table fit strategy. Keep internal identifiers out of reader-facing prose unless they are mapped once for the reader and then moved back out of prose, and record the terminology-clarity self-check, the section-level acceptance gate, section-style policy compliance, the protocol/scope impact audit, the export or remote bundle audit, the round target layer, any canonical-only justification while workflow-language was active, any cross-language sync justification, the active canonical/workflow-language roots, the resolved target path role, any out-of-band justification, and the table-semantics audit in the write iteration artifact. If the manuscript would start from the managed scaffold and no template decision is recorded yet, ask once whether to keep the default scaffold or attach a template directory first. If finalization reaches a round where `workflow_language` and `paper_language` differ, finish and preserve the workflow-language paper layer first, then ask once whether to keep the draft language or convert the canonical manuscript to `paper_language`, persist that answer, record both the language decision and the workflow-language paper-layer path in the latest write iteration, and only then edit the final manuscript in the chosen language.

package/package-assets/shared/lab/.managed/scripts/validate_section_draft.py

@@ -4,6 +4,12 @@ import json
 import re
 import sys
 from pathlib import Path
+from paper_topology import (
+    classify_paper_path,
+    find_project_root,
+    load_workflow_config,
+    resolve_paper_topology,
+)
 from validate_rule_preflight import validate_rule_preflight
 
 
@@ -60,26 +66,6 @@ def has_local_naming_bridge(text: str) -> bool:
     )
 
 
-def find_project_root(section_path: Path) -> Path | None:
-    for parent in (section_path.parent, *section_path.parents):
-        workflow_config = parent / ".lab" / "config" / "workflow.json"
-        if workflow_config.exists():
-            return parent
-    return None
-
-
-def load_workflow_config(project_root: Path | None) -> dict:
-    if project_root is None:
-        return {}
-    workflow_config = project_root / ".lab" / "config" / "workflow.json"
-    if not workflow_config.exists():
-        return {}
-    try:
-        return json.loads(workflow_config.read_text(encoding="utf-8"))
-    except json.JSONDecodeError:
-        return {}
-
-
 def extract_markdown_field(text: str, section_heading: str, field_label: str) -> str:
     section_match = re.search(
         rf"^##\s+{re.escape(section_heading)}\s*$([\s\S]*?)(?=^##\s+|\Z)",
@@ -317,6 +303,53 @@ def check_neighbor_asset_files(section: str, section_path: Path, issues: list[st
             )
 
 
+def check_paper_topology_targeting(section_path: Path, issues: list[str]):
+    project_root = find_project_root(section_path)
+    workflow_config = load_workflow_config(project_root)
+    if project_root is None or not workflow_config:
+        return
+
+    topology = resolve_paper_topology(project_root, workflow_config)
+    target_role, _ = classify_paper_path(section_path, topology)
+    if target_role in {"active-canonical", "active-workflow-language"}:
+        return
+
+    iteration_dir = project_root / ".lab" / "writing" / "iterations"
+    if not iteration_dir.exists():
+        issues.append(
+            "section target is outside the active paper topology, but no write-iteration audit explains why this out-of-band edit was acceptable"
+        )
+        return
+
+    iteration_files = sorted(iteration_dir.glob("*.md"))
+    if not iteration_files:
+        issues.append(
+            "section target is outside the active paper topology, but no write-iteration audit explains why this out-of-band edit was acceptable"
+        )
+        return
+
+    iteration_text = read_text(iteration_files[-1])
+    recorded_role = extract_markdown_field(
+        iteration_text,
+        "Paper Topology Target",
+        "Resolved target path role:",
+    )
+    out_of_band_reason = extract_markdown_field(
+        iteration_text,
+        "Paper Topology Target",
+        "If the target path was outside the active paper topology, why was out-of-band editing acceptable:",
+    )
+
+    if recorded_role and target_role not in recorded_role:
+        issues.append(
+            f"section target is outside the active paper topology, but the latest write iteration records a different target role ({recorded_role})"
+        )
+    if not has_meaningful_field_value(out_of_band_reason):
+        issues.append(
+            "section target is outside the active paper topology, but the latest write iteration does not explain why this out-of-band edit was acceptable"
+        )
+
+
 def check_workflow_language_targeting(section_path: Path, issues: list[str]):
     project_root = find_project_root(section_path)
     workflow_config = load_workflow_config(project_root)
@@ -325,7 +358,10 @@ def check_workflow_language_targeting(section_path: Path, issues: list[str]):
     if not workflow_language or not paper_language or workflow_language == paper_language:
         return
 
-    if "workflow-language" in section_path.parts:
+    topology = resolve_paper_topology(project_root, workflow_config) if project_root else None
+    target_role = classify_paper_path(section_path, topology)[0] if topology else ""
+
+    if target_role != "active-canonical":
         return
 
     if project_root is None:
@@ -517,6 +553,7 @@ def main():
     issues: list[str] = []
     check_common_section_gate_risks(text, issues)
     check_write_rule_preflight(section_path, issues)
+    check_paper_topology_targeting(section_path, issues)
     check_workflow_language_targeting(section_path, issues)
     check_section_style_policy(text, args.section, issues)
     SECTION_CHECKS[args.section](text, issues)

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

@@ -92,6 +92,13 @@
 - If both canonical manuscript and workflow-language layer were edited, why:
 - Was cross-language sync explicitly requested or required by final-draft/export:
 
+## Paper Topology Target
+
+- Active canonical paper root:
+- Active workflow-language paper root:
+- Resolved target path role:
+- If the target path was outside the active paper topology, why was out-of-band editing acceptable:
+
 ## Protocol / Scope Impact Audit
 
 - Did this round replace a canonical experiment or evaluation protocol:

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

@@ -225,8 +225,10 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - Ordinary manuscript drafting rounds should follow `workflow_language`.
 - In those ordinary rounds, the manuscript is still LaTeX, so the initial `.tex` section drafts must stay in `workflow_language` rather than jumping straight to `paper_language`.
 - When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as active and use it as the default ordinary working layer.
+- Resolve the active paper topology from `.lab/config/workflow.json` before drafting; only paths under `<deliverables_root>/paper/` or `<deliverables_root>/paper/workflow-language/` count as managed manuscript targets.
 - Ordinary write rounds should still edit one target paper layer at a time, not both language layers at once.
 - If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization.
+- If the named file falls outside the active paper topology, treat it as an out-of-band file edit instead of a managed manuscript round.
 - If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the latest write iteration artifact.
 - If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active.
 - Treat the workflow-language paper as a real persisted paper layer, not as a review layer.
@@ -279,6 +281,7 @@ Use this skill when the user invokes `/lab:*` or asks for the structured researc
 - Record the round target layer in the latest write iteration artifact as `canonical manuscript`, `workflow-language paper layer`, or `both`.
 - If workflow-language was active and the round still targeted the canonical manuscript, record why canonical-only writing was acceptable in the latest write iteration artifact.
 - If both layers were edited, record why the cross-language sync was required and whether it was explicitly requested by the user or required by final-draft/export finalization.
+- Record the active canonical paper root, active workflow-language paper root, resolved target path role, and any out-of-band justification in the latest write iteration artifact.
 - Record the protocol/scope impact audit in the latest write iteration artifact whenever a round replaces or rewrites a canonical experiment/evaluation protocol.
 - If the round is an export or remote-publication round, record the export target, whether the workflow-language paper layer was included in the exported or pushed bundle, and why any canonical-only export was acceptable.
 - If a claim is not supported by evidence, weaken or remove it.

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

@@ -75,8 +75,11 @@ Run these on every round:
 - Ordinary manuscript drafting rounds should follow `workflow_language`.
 - During ordinary rounds, ordinary `.tex` section drafts must stay in `workflow_language`; do not treat `paper_language` as the default draft language.
 - When `workflow_language` and `paper_language` differ, treat the workflow-language paper layer as active and use it as the default ordinary working layer.
+- Resolve the active paper topology from `.lab/config/workflow.json` before drafting: the active canonical layer is `<deliverables_root>/paper/`, and when `workflow_language` differs from `paper_language` the active workflow-language layer is `<deliverables_root>/paper/workflow-language/`.
 - Ordinary write rounds should still edit one target paper layer at a time, not both language layers at once.
 - If the user names a concrete file or layer, treat that as the only target for the round unless they also explicitly request synchronization.
+- Classify the named target path before editing it. Active-layer targets (`active-canonical` / `active-workflow-language`) remain managed manuscript rounds; deprecated or legacy paper-layer targets are only acceptable as out-of-band file edits and must not silently replace the active paper topology.
+- Do not treat old layer names such as `review_zh`, `translation_zh`, `sections_zh`, or stale `deliverables/.../workflow-language/*.md` paths as active manuscript roots, even if older project files still reference them.
 - If a workflow-language paper layer is active and the round still targets the canonical manuscript, record why canonical-only writing was acceptable in the write-iteration artifact instead of silently treating canonical as the default working layer.
 - If `paper_language_finalization_decision=convert-to-paper-language`, explicit canonical-manuscript work may target the canonical `paper_language` manuscript, but that does not make canonical the default ordinary working layer while workflow-language remains active.
 - Only edit both the canonical manuscript and the workflow-language paper layer in the same round when the user explicitly asks for cross-language synchronization or when a final-draft/export language-finalization step requires both layers to be refreshed together.
@@ -153,6 +156,7 @@ Run these on every round:
   - record what each figure or analysis asset should show and why the reader needs it
   - record which citation anchors must appear in the section and why each anchor matters
 - Before drafting `introduction`, `method`, `experiments`, `related work`, or `conclusion`, run `.lab/.managed/scripts/validate_paper_plan.py --paper-plan .lab/writing/plan.md`.
+- Before drafting `introduction`, `method`, `experiments`, `related work`, or `conclusion`, also run `.lab/.managed/scripts/validate_paper_topology.py --project-root .` so plan/context files cannot keep presenting legacy layers as the active paper topology.
 - When the repository workflow config is available, the paper-plan validator also checks that `.lab/writing/plan.md` stays in `workflow_language` instead of silently drifting into another language.
 - If the paper-plan validator fails, stop and fill `.lab/writing/plan.md` first instead of drafting prose.
 - During ordinary draft rounds, run `.lab/.managed/scripts/validate_section_draft.py --section <section> --section-file <section-file> --mode draft` and `.lab/.managed/scripts/validate_paper_claims.py --section-file <section-file> --mode draft` after revising the active section.
@@ -198,6 +202,7 @@ Run these on every round:
 - Record the round target layer in the write-iteration artifact as `canonical manuscript`, `workflow-language paper layer`, or `both`.
 - If workflow-language was active and the round still targeted the canonical manuscript, record why canonical-only writing was acceptable in the write-iteration artifact.
 - If both layers were edited, record why the cross-language sync was required and whether it was explicitly requested by the user or required by final-draft/export finalization.
+- Record the active canonical paper root, the active workflow-language paper root when present, the resolved target path role, and any out-of-band reason in the write-iteration artifact before accepting the round.
 - Record the protocol/scope impact audit in the write-iteration artifact whenever a round replaces or rewrites a canonical experiment/evaluation protocol.
 - If the round is an export or remote-publication round, record the export target, whether the workflow-language paper layer was included in the exported or pushed bundle, and why any canonical-only export was acceptable.
 - 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.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "superlab",
-  "version": "0.1.62",
+  "version": "0.1.63",
   "description": "Strict /lab research workflow installer for Codex and Claude",
   "keywords": [
     "codex",