cnki-paper-detail

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass, no license.

Audit 0.4.1 · source sha 84616505ce85 · 26 Oct 2026Report an issue or request a re-scan

• Publishercookjohn GitHub userOwner of the repository this skill is read from
• Source foldercookjohn/cnki-skills/skills/cnki-paper-detailThe exact folder we audited
• LicenceNot declaredAs the repository states it
• Commit pinned for this verdict84616505ce85A new audit runs when the source changes
• Repository statusNot archivedLast push 226 days ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Extract full paper details from a CNKI paper page including title, authors, affiliations, abstract, keywords, fund, classification. Use when the user needs detailed information about a specific paper.

---
name: cnki-paper-detail
description: Extract full paper details from a CNKI paper page including title, authors, affiliations, abstract, keywords, fund, classification. Use when the user needs detailed information about a specific paper.
argument-hint: "[paper URL or blank if already on detail page]"
---

# CNKI Paper Detail Extraction

Extract complete metadata from a CNKI paper detail page.

## Arguments

`$ARGUMENTS` is optionally a CNKI paper detail URL (containing `kcms2/article/abstract`). If not provided, assumes the current page is already a paper detail page.

## Steps

### 1. Navigate to the paper page (if URL provided)

If `$ARGUMENTS` contains a URL:
- Use `mcp__chrome-devtools__navigate_page` with the URL.
- Use `mcp__chrome-devtools__wait_for` with text `["摘要"]` and timeout 15000.

### 2. Check for captcha

Use `mcp__chrome-devtools__take_snapshot`. If "拖动下方拼图完成验证" found, notify user:

> CNKI 正在显示滑块验证码。请在 Chrome 浏览器中手动完成拼图验证,完成后告诉我继续。

### 3. Extract paper metadata via JavaScript

Use `mcp__chrome-devtools__evaluate_script` with this function:

```javascript
() => {
  const brief = document.querySelector('.brief');
  if (!brief) return { error: 'Paper detail section (.brief) not found' };

  // Title
  const title = brief.querySelector('h1')?.innerText?.trim()
    ?.replace(/\s*附视频\s*$/, '')  // remove "附视频" suffix
    ?.replace(/\s*网络首发\s*$/, ''); // remove "网络首发" suffix

  // Authors - first h3.author contains author links with sup tags
  const authorH3s = brief.querySelectorAll('h3.author');
  const authorSection = authorH3s[0];
  const authors = [];
  if (authorSection) {
    const authorLinks = authorSection.querySelectorAll('a');
    authorLinks.forEach(a => {
      const name = a.innerText?.replace(/\d+$/, '').trim();
      const supMatch = a.innerText?.match(/(\d+)$/);
      const affiliationNum = supMatch ? supMatch[1] : '';
      authors.push({ name, affiliationNum });
    });
  }

  // Affiliations - second h3.author contains org links
  const affiliations = [];
  if (authorH3s.length > 1) {
    const orgLinks = authorH3s[1].querySelectorAll('a');
    orgLinks.forEach(a => {
      affiliations.push(a.innerText?.trim());
    });
  }

  // Abstract
  const abstractEl = document.querySelector('.abstract-text');
  const abstract = abstractEl?.innerText?.trim() || '';

  // Keywords
  const keywordsP = document.querySelector('p.keywords');
  const keywords = keywordsP
    ? Array.from(keywordsP.querySelectorAll('a')).map(a => a.innerText?.replace(/;$/, '').trim())
    : [];

  // Fund
  const fundsP = document.querySelector('p.funds');
  const fund = fundsP?.innerText?.trim() || '';

  // Classification code
  const clcCode = document.querySelector('.clc-code');
  const classification = clcCode?.innerText?.trim() || '';

  // Journal/source
  const docTop = document.querySelector('.doc-top');
  const journal = docTop?.querySelector('a')?.innerText?.trim() || '';

  // Online first / publication info
  const headTime = document.querySelector('.head-time');
  const pubInfo = headTime?.innerText?.trim() || '';

  // Is online first?
  const isOnlineFirst = !!brief.querySelector('.icon-shoufa');

  // Article outline/TOC
  const catalogList = document.querySelector('.catalog-list, .catalog-listDiv');
  const toc = catalogList?.innerText?.trim() || '';

  // Citation network counts
  const citationTabs = document.querySelectorAll('ul.module-tab.tpl_lieteratures li');
  const citationInfo = {};
  citationTabs.forEach(li => {
    const id = li.getAttribute('data-id');
    const text = li.innerText?.trim();
    const countMatch = text.match(/(\d+)/);
    if (id) {
      citationInfo[id] = {
        label: text.replace(/\d+/, '').trim(),
        count: countMatch ? parseInt(countMatch[1]) : 0
      };
    }
  });

  return {
    title,
    authors,
    affiliations,
    abstract,
    keywords,
    fund,
    classification,
    journal,
    pubInfo,
    isOnlineFirst,
    toc,
    citationInfo
  };
}
```

### 4. Format and present the output

```
## {title} {isOnlineFirst ? "[网络首发]" : ""}

**Authors:**
{For each author: "- {name} ({affiliation})"}

**Affiliations:**
{For each affiliation: "- {affiliation}"}

**Journal:** {journal}
**Publication Info:** {pubInfo}

**Abstract:**
{abstract}

**Keywords:** {keywords joined by ", "}

**Fund:** {fund}
**Classification:** {classification}

**Citation Network:**
{For each citation type: "- {label}: {count}"}
```

### 5. Fallback: snapshot-based parsing

If JS extraction fails, use `mcp__chrome-devtools__take_snapshot` and parse the accessibility tree:
- **Title**: `heading` level 1 element
- **Authors**: `link` elements whose URLs contain `kcms2/author/detail`
- **Affiliations**: `link` elements whose URLs contain `kcms2/organ/detail`
- **Abstract**: `StaticText` following "摘要:"
- **Keywords**: `link` elements whose URLs contain `kcms2/keyword/detail`
- **Fund**: `link` elements following "基金资助:"
- **Classification**: `StaticText` following "分类号:"

## Verified DOM Selectors

| Data           | Selector                                   | Notes                                    |
|----------------|--------------------------------------------|------------------------------------------|
| Paper section  | `.brief`                                    | Main paper info container                |
| Title          | `.brief h1`                                 | May contain icons, clean text needed     |
| Authors        | `.brief h3.author:first-of-type a`          | Text has superscript numbers (e.g., "张三1") |
| Affiliations   | `.brief h3.author:nth-of-type(2) a`         | Text starts with "N." (e.g., "1.北京大学") |
| Abstract       | `.abstract-text`                            | Full abstract text                       |
| Keywords       | `p.keywords a`                              | Semicolon-separated keyword links        |
| Fund           | `p.funds`                                   | Fund information text                    |
| Cla

Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/cookjohn/cnki-skills
git -C cnki-skills checkout 84616505ce85

Works in

1. 2026-10-26
   ✓SAFE B 89 / 100

   first audit · source 84616505ce85

Questions about cnki-paper-detail

• github.com/cookjohn/cnki-skills/tree/master/skills/cnki-paper-de — the audited source at commit 84616505ce85 and every finding aboveObserved · 26 Oct 2026

Spotted something wrong? Report a correction · Methodology