Menu
Proposal-first scientific writing pipeline. Three modes (compose/revise/hybrid) with four-layer QA pipeline. Enforces evidence-before-prose, argument-before-sections, and contracts-before-paragraphs.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Use this skill for managed MLFF molecular dynamics, restart-safe trajectory continuation, ensemble selection, and trajectory-health analysis when mlff_md is available.
Use only after a DPDispatcher-backed managed execution tool returns a failure with receipt/context fields, an ambiguous transport error, or evidence of a possible orphan job. Do not use for a pending synchronous call or ordinary success.
Use this skill before remote_submission or remote_submission_batch to build and verify canonical stage directories for registered DPDispatcher tasks, including deterministic MLFF SP/relax, MD, and NEB input layouts.
Use this skill for managed MLFF NEB optimization after a complete locally interpolated fixed-image path has been validated.
Use this skill for MLFF single-point screening, ranking, and geometry relaxation when choosing among enabled MACE, FairChem UMA, MatterSim, or ORB-v3 backends.
Use this skill for dispatching prepared VASP jobs with remote_submission or remote_submission_batch, choosing valid stage layouts, and collecting clean failure evidence.
| name | researchwrite |
| description | Proposal-first scientific writing pipeline. Three modes (compose/revise/hybrid) with four-layer QA pipeline. Enforces evidence-before-prose, argument-before-sections, and contracts-before-paragraphs. |
| version | 1.0.0 |
| author | 十五 (JL Lab) |
| license | MIT |
| metadata | {"hermes":{"tags":["research","writing","proposal","revision","qa","multi-agent"],"related_skills":["brainstorming","professor","humanizer","docx"]}} |
受 autonovel（状态机+打分）、professor（动态专家）、brainstorming（入口追问）、humanizer（语言清理）启发的科研写作状态机。不是通用"帮我写论文"prompt。
research_canon 和 evidence_tableargument_mapprofessor 按失败模式召唤对应专家| 输入 | 模式 |
|---|---|
| 题目、方向、模糊想法 | compose — 加载 references/compose-mode.md |
| 已有段落/章节/完整 proposal | revise — 加载 references/revise-mode.md |
| 已有草稿 + 扩写/补写/重构 | hybrid — 加载 references/hybrid-mode.md |
模糊时推断默认模式。只在选择会改变工作流时才问用户。
工作目录：<outputs>/researchwrite/<project-slug>/
标准文件：
00_scope.md 写作任务边界
01_research_canon.md 硬事实和约束
02_evidence_table.md claim → evidence 映射表
03_argument_map.md 论证架构
04_section_contracts.md 每节的 purpose / inputs / allowed & forbidden claims
05_style_guide.md 风格、术语、禁用表达
state.json 项目状态（mode、round、scores、technical_debts）
sources/ 用户材料、文献、数据
drafts/ 草稿和分节文件
revision_briefs/ 修订简报
qa_logs/ 诊断、专家审查、anti-slop、打分记录
exports/ 最终输出（.md + .docx）
新建项目时从 templates/ 取空模板。references/worked-example-quaternary-proposal.md 提供了基于通用材料科学领域的完整填写示例。
公开版本包含四个关键 reference：
| Reference | 用途 |
|---|---|
references/evaluation-rubric.md | 8 维 × 4 锚点评分体系 + 评分流程 |
references/stopping-rules.md | 迭代循环终止条件 |
references/foundation-files.md | 建立 foundation 五文件 |
references/validation-checklist.md | 自动校验清单 |
完整 reference 文件（compose/revise/hybrid 模式详解、专家分派、导出归档、综述框架等 12 份）未包含在公开版本中。如需获取，请通过 GitHub issue 或邮件联系作者。
每次运行结束时输出：
当用户说"审查这段/跑 QA/检查方案/过 pipeline"时，进入 QA 模式。
| 挡位 | 适用场景 | 阈值 |
|---|---|---|
paper | 投稿论文 | 7.0 |
proposal | 研究方案/开题 | 7.0 |
internal | 内部汇报/周报 | 5.0 |
quick | 快速扫读 | 无 |
Gate 2: professor Convener（内容层）
├── 论文 → 方法论专家 + 领域专家
├── proposal → 可行性专家 + 创新性专家
└── 文献reviews → 覆盖面专家 + 批判深度专家
│
▼
Gate 1: humanizer 语言审读
└── 按原版 humanizer 规则检查和修改文风
│
▼
Gate 3: auto-validation（格式/完整性层）
└── 每个 claim 有 citation？方法可复现？编号连续？
│
▼
Gate 4: 评分阈值（分维度打分）
├── 总分≥阈值 → 通过 ✅
└── 总分<阈值 → 按低分维度定向回退（不超过 3 轮）
Gate 2 在 Gate 1 之前 — 避免改了句子后被专家打回重写。
用户: "用 researchwrite 审查这段 discussion，paper 挡位"
→ 自动走 Gate 2 → Gate 1 → Gate 3 → Gate 4
→ 返回审查报告 + 定向修改建议
用户: "快速扫一下这个邮件"
→ 走 quick 挡位，只标记 P0 问题
首次使用告诉 agent 你的研究背景，agent 会调用 professor 建立领域专家知识。后续写作中专家审查会基于你的领域。