csdn-publish-and-data

name	csdn-publish-and-data
description	Save Markdown articles into the CSDN draft box and fetch CSDN creator analytics/article performance data with storageState.json or cookies.json. Use when the user asks to 保存 CSDN 草稿, 准备 CSDN 文章草稿, 批量生成 CSDN 草稿, 抓取 CSDN 创作中心运营数据, 获取最近文章的 7 日动态数据, 校验 CSDN 创作中心登录态, or export CSDN creator storage state.

CSDN Publish And Data

Language

Match the user's language.

Use This Skill For

• Save a local Markdown file into the CSDN draft box.
• Validate whether the current CSDN creator auth state is still usable.
• Export Playwright storage state for later reuse.
• Fetch creator analytics with a JSON report focused on overall metrics plus per-article summary and recent 7-day dynamics.

Do not use this skill for password-based login, CAPTCHA bypass, multi-account automation, or hidden-field guessing.

Resolve Paths And Runtime

Determine this SKILL.md directory as {baseDir}. Use scripts from {baseDir}/scripts.

Resolve ${RUN_TS} like this:

1. Use npx tsx if dependencies are already installed.
2. Run npm install inside {baseDir}/scripts only when dependencies are missing.
3. Ask the user to install Node.js LTS only when Node.js is unavailable.

Choose The Workflow

Pick exactly one path for the current request:

• Post article (draft only): use post-article.ts.
• Fetch analytics: use fetch-analytics.ts.
• Check auth only: use check-login.ts.
• Export auth state only: use export-storage-state.ts.

Run one-time setup steps only when the environment is new, dependencies are missing, browser support changed, or the saved auth file is missing/expired.

One-Time Setup

Run this checklist only for first-time setup or when the previous setup is no longer valid.

One-Time Setup Progress:
- [ ] Step 1: Install dependencies in {baseDir}/scripts
- [ ] Step 2: Install Playwright browser if needed
- [ ] Step 3: Export or place storageState.json / cookies.json
- [ ] Step 4: Run login check once to confirm auth works

Useful commands:

cd {baseDir}/scripts
npm install
npx playwright install chromium
npx tsx check-login.ts --state <storageState.json>

After this succeeds once, skip these steps in normal runs unless auth expires or the machine/runtime changes.

Auth File Discovery Order

1. CLI --state path
2. CLI --cookie path
3. /.auth/storageState.json
4. /.auth/cookies.json

Prefer storageState.json. Use cookies.json only as fallback.

Read extra references only when needed:

• First-time config and defaults: references/config/first-time-setup.md
• Export auth state: references/auth/export-storage-state.md
• Draft-saving details: references/article-posting.md
• Output schema details: references/output-format.md
• Troubleshooting: references/troubleshooting/common-issues.md
• Ubuntu headless setup: references/ubuntu/headless-setup.md

Recurring Workflow: Save Draft

Use this path for requests like “把这篇文章保存到 CSDN 草稿箱”.

Draft Workflow Progress:
- [ ] Step 1: Resolve auth file
- [ ] Step 2: Run post-article.ts with the original Markdown file
- [ ] Step 3: Confirm draft save result from JSON output
- [ ] Step 4: Report draft result files and warnings

Recommended command:

npx tsx {baseDir}/scripts/post-article.ts --file <article.md> --draft --state <storageState.json>

Rules:

• Pass the original Markdown file directly.
• Do not convert Markdown to HTML before calling the script.
• By default, saves draft only.
• If the editor layout changes or selectors fail, rerun with --headful and inspect the current page.
• If auth is already known-good, do not rerun dependency install or login check first.

Recurring Workflow: Fetch Analytics

Use this path for requests like “抓取 CSDN 运营数据”, “获取最近文章 7 日数据”, or “导出 CSDN 创作中心表现”.

Analytics Workflow Progress:
- [ ] Step 1: Resolve auth file
- [ ] Step 2: Run fetch-analytics.ts
- [ ] Step 3: Return the main JSON report path
- [ ] Step 4: Return raw output path only if --save-raw was used or debugging is needed

Recommended command:

npx tsx {baseDir}/scripts/fetch-analytics.ts --page both --state <storageState.json> --save-raw --output <output-dir>

Default interpretation of the analytics output:

• Main JSON is the primary report for analysis.
• Main JSON keeps only analysis-relevant fields: overall creator metrics plus per-article summary and recent 7-day dynamics.
• Raw files are for debugging and schema inspection only.
• If auth is already known-good, do not rerun dependency install or login check first.

Current report shape:

{
  "generatedAt": "...",
  "report": {
    "range": { "start": "YYYY-MM-DD", "end": "YYYY-MM-DD" },
    "overview": {
      "articleCount": 0,
      "exposuresCount": 0,
      "diggCount": 0,
      "commentCount": 0,
      "viewCount": 0,
      "score": 0,
      "averageQuality": 0,
      "collectCount": 0
    },
    "articles": [
      {
        "articleId": "...",
        "title": "...",
        "url": "...",
        "summary": {
          "publishTime": "...",
          "exposuresCount": 0,
          "viewCount": 0,
          "commentCount": 0,
          "diggCount": 0,
          "favoriteCount": 0,
          "fansCount": 0,
          "qualityScore": {
            "score": 0,
            "version": "V5.0",
            "createAt": "..."
          }
        },
        "recent7Days": [
          {
            "date": "YYYY-MM-DD",
            "exposuresCount": 0,
            "viewCount": 0,
            "commentCount": 0,
            "diggCount": 0,
            "favoriteCount": 0,
            "fansCount": 0
          }
        ]
      }
    ]
  }
}

Common options:

• --page analytics|manage|both
• --start YYYY-MM-DD
• --end YYYY-MM-DD
• --output <dir>
• --state <path>
• --cookie <path>
• --probe
• --headful
• --timeout <ms>

Focused Execution Rules

• Prefer the main JSON report for user-facing answers.
• Mention raw files only when debugging, verification, or schema exploration is useful.
• Treat storage state as reusable session state; do not force refresh unless login has expired.
• Keep the user-facing summary focused on meaningful analytics rather than raw transport metadata.

Auth-Only Workflow

Use this only when the request is specifically about validating or exporting auth state.

Check auth:

npx tsx {baseDir}/scripts/check-login.ts --state <storageState.json>

Export storage state:

npx tsx {baseDir}/scripts/export-storage-state.ts

Safety And Boundaries

• Do not ask for account password, SMS code, or login QR secrets.
• Do not commit cookies.json or storageState.json into git.
• Draft save is always supported.
• If CSDN introduces new required fields in the editor or publish dialog, stop after surfacing the missing selector or warning instead of guessing hidden values.