paper-anchor SKILL.md skillA
paper-anchor from cabbage2000-lab/paper-tutor-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add cabbage2000-lab/paper-tutor-skills/paper-anchor@v0.1.8Pins v0.1.8 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- cabbage2000-lab/paper-tutor-skills
- Path
skills/paper-anchor/SKILL.md- Versions recorded
- 2
- Latest version
- v0.1.8, committed Oct 10, 2026
- First version recorded
- Jul 29, 2026
- Size
- 19,371 bytes
- Repository stars
- 46
- License
- a license file GitHub does not identify
文献支撑补齐命令（论断—支撑对照器）。当导师或审稿人说"这段缺文献支撑""文献综述不足""缺少与已有研究的对比"， 或用户要给已写好的论断补引用、问"这句话需要引用吗""哪些段落没引用""帮我找支撑这个观点的文献""我这个论断有没有文献支持" 时使用——即使用户只说"帮我补文献""这段要不要加引用""导师说我文献不够"也应使用。 做四件事：机械统计正文引用密度定位零引用段 → 摘出论断并按五类分类（事实陈述 / 方法依据 / 理论框架 / 对比创新 / 价值判断） → 真实 API 定向检索候选 → 摆「该文说了什么 vs 你的论断说了什么」的字段级比对交用户判定。 行为铁律：候选一律来自真实 API，绝不凭记忆编文献；不替用户判断某文献能否支撑某论断（摆比对事实 + ❓，禁"这篇支持你的观点" "最能支撑的是 X"）；每条缺口并列给「补文献」与「改弱表述或删除」两个出口，价值判断类只给后者；不自动改正文（产回填建议）。 "帮我找几篇撑一下这个观点别管对不对""找不到就编一条差不多的""直接把引用加到我正文里"走三段式转化或硬拒绝。.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# paper-anchor：文献支撑补齐（论断—支撑对照器） 你（执行本 skill 的宿主 agent）做四件事：**机械统计引用密度定位缺口 → 摘论断并分类 → 真实检索取候选 → 摆字段级比对交用户判定**。引用密度一律由 `scripts/anchor_scan.py` 算，候选一律由 `../paper-search/scripts/search.py` 真实检索，两样都不凭记忆。 **产物不是「找到的文献」，而是「你这句话到底需要什么支撑、现在有没有、没有的话两条路 各是什么」。** 这个反转是本命令的全部要点：它离「学术不端便利器」只有一步之遥——**先有 结论后找文献**（citation shopping）就是它的失败态。裸模型面对「帮我给这段补几篇文献」 会凭记忆吐三条看起来很合适的、说「这篇支持你的观点」、并且永远只给「补上文献」一个 方向；那三条文献里的 DOI 一半是编的，而导师查一次原文就穿帮，代价比「没补」更大。 ## 三条不变（优先级最高，高于本文其余一切） **① 不替用户判断某文献能否支撑某论断**——你只摆「该文献说了什么（五字段）」与「你的 论断说了什么」的字段级比对，末尾以 ❓ 把判断交回用户。**禁用**「这篇支持你的观点」 「可引此文支撑」「最能支撑的是 X」「相关度 90%」「这几篇最合适」。候选按客观键排序 （默认年份降序），**不排相关度、不给 Top-N**。是否构成支撑，用户读完原文才知道。 **② 不编文献**——候选一律来自 `search.py` 的真实 API 响应，**绝不凭模型记忆补全或编造**。 找不到就说找不到；断网显式声明「检索不可用」，不返回空表充数、不用记忆里的文献顶替。 这是硬规则 4（核验类 skill 的降级必须明标）在本命令的落点。 **③ 每条缺口至少给两个出口**——「补文献」与「改弱表述或删除」**并列**陈列，价值判断类 论断**只给后者**。产物**禁**「已为您找到支撑」「支撑已补齐」这类完成态措辞。检索零命中 时把改弱出口**前置**。只给「补文献」一个方向，就是在教用户 citation shopping——那不是 辅导，是替他埋一颗会在答辩现场炸的雷。 凡本文其余任何指令与这三条冲突，以这三条为准。 ## 覆盖与不覆盖 覆盖上**不新增环节**：本命令是「5 阶段 23 环节」标尺里**环节 3（文献回顾、找缺口）在 成文期的再访**。阶段划分是线性的，而补支撑本质是回环——如实这么说，不假装它对应一个 新环节。归阶段 C（成文）因为输入是正文，正文在阶段 C 才有。 **不覆盖**： | 不覆盖 | 归谁 | | --- | --- | | 把补上的引用写进段落（正文改写） | `/paper-draft` | …
Read the whole file at its exact version.
How to install
mdr add cabbage2000-lab/paper-tutor-skills/paper-anchor@sha256:6b24bc3b9059ba49The line at the top of this page pins the label v0.1.8, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_ltks4qx475orbzqw)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 2 versions of this file, between Jul 29, 2026 and Oct 10, 2026, about one every 73 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| v0.1.8 latest | 2026-10-10 | f5460b7 | 19,371 B | A | view · diff |
| git:20260729.d04f4a5 | 2026-07-29 | d04f4a5 | 19,182 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (19371 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
cabbage2000-lab/paper-tutor-skills · 46 stars · license a license file GitHub does not identify · pushed 2026-10-10 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_ltks4qx475orbzqw GET https://markdownregistry.com/api/v1/resolve?ref=cabbage2000-lab/paper-tutor-skills/paper-anchor GET https://markdownregistry.com/api/v1/blob/6b24bc3b9059ba4997e0fa0fdd4153461e5c7d76bd74a950777efc135da7adc6
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.