xiaohongshu-account-analyzer

Xiaohongshu account in-depth diagnosis

Task objective

• This Skill is used to: deeply deconstruct Xiaohongshu accounts and output a professional commercial value analysis report
• Capabilities include: seven-dimension scoring (account positioning 10 points, follower profile and needs 15 points, topic selection system 10 points, cover style 10 points, viral post capability 20 points, interaction scale 20 points, update capacity 15 points, full score 100), action prescription generation, similar account recommendations, multi-account comparison analysis
• Trigger conditions: the user provides a Xiaohongshu ID (the Xiaohongshu ID below the name, in the format of pure numbers or an alphanumeric combination, not a Chinese nickname), and a complete account diagnostic report needs to be generated

Usage examples

Scenario	User input	Expected output
Single account diagnosis	Analyze Xiaohongshu ID 26112666886	Output a seven-dimension diagnostic report, recommend similar accounts, and generate an HTML report that can be exported as images
Multi-account comparison	Help me compare and analyze ID1 and ID2	Output diagnostic reports for each account separately, generate a horizontal comparison summary, and generate a comparison HTML report
Analyze similar accounts	Reply with number "1" to select a similar account	Perform a complete diagnostic analysis on the selected similar account, output a report, and generate HTML

🔑 Authentication

Get API Key

Please go to Redfox Hub to get the API KEY.

Configure API Key

Set REDFOX_API_KEY as an environment variable:

export REDFOX_API_KEY="ak_xxxx..."

The script reads the key through os.getenv('REDFOX_API_KEY') , and when it is missing, it errors out and provides configuration guidance. Request headers automatically carry X-API-KEY。

---

Prerequisites

• Data interface: POST https://redfox.hk/story/api/xhsUser/query（详见 references/api_guide.md）
• WebSearch: after obtaining the blogger's nickname, perform background information completion
• Scoring rules and data filling specifications: see references/workflow_guide.md

Operation steps

Standard process (single account diagnosis)

Step 1: Opening guidance

• Output references/workflow_guide.md the standard opening in
• Guide the user to enter a Xiaohongshu ID (pure numbers or an alphanumeric combination, not a Chinese nickname)

Step 2: Data query

• Identify user input: pure numbers or an alphanumeric combination are recognized as a Xiaohongshu ID
• Call script:python scripts/xiaohongshu_analyzer.py query --user_ids "小红书号"
• The script saves the raw data to output/raw_data.json
• If the user enters a Chinese nickname, prompt the user to provide the Xiaohongshu ID

Step 3: WebSearch background completion

• Search "Xiaohongshu + nickname" to complete background information
• Search "nickname + interview/report" to find media coverage
• Search for cross-platform presence such as Douyin/Bilibili/WeChat Official Account

Step 4: Query result handling

• Failure threshold(return query_type: "threshold_exceeded"）：

  当前小红书号已超过失败阈值，请联系客服邮箱redfoxdata@proton.me 或访问 https://redfox.hk/?source=github处理
  

• Account not found(account data is empty):

  未查询到相关账号：当前 Skill 仅收录热门账号。如需定制数据，可邮件联系[红狐数据](https://redfox.hk/?source=github)咨询：redfoxdata@proton.me
  

• Query successful → continue to steps 5-7
• Multi-account data → switch to the multi-account comparison process

Step 5: Diagnostic report generation

• Read references/report_template.md to get the report format
• Read references/workflow_guide.md to get the scoring rules and data entry specifications
• Fill the script data into the template and output the diagnostic report directly in the conversation (do not generate an md file)
• Account works module prompt (must output): Before displaying any Xiaohongshu work link, you must first output:

  ！！！受小红书风控规则限制，部分作品链接可能无法正常跳转，您可复制对应作品标题前往小红书搜索查看，感谢理解🙇‍♀️🙇‍♀️
  

• Data field description: gmtCreate is the time when the Honghu data system first included this account, not the Xiaohongshu account registration time, and must not be used to judge the account operation duration
• Comprehensive score conclusion rules：

  • =60 points: list >=2 verified and reusable specific actions

  • <60 points: first point out the data problem corresponding to the weak item, then give suggestions
• Empty value handling: When a data field is empty, directly hide the corresponding module and do not display "No data"

Step 6: Display similar accounts

• From the data returned by the script, use the similar_accounts field to extract 2-5 similar accounts
• Display directly in the format (no need to ask), and after display is complete, proceed to step 7

Step 7: Generate HTML report

• Step 7.1: Generate data template
  • python scripts/xiaohongshu_analyzer.py build_report_data --account_name "账号名"
• Step 7.2: Fill AI analysis content into output/report_data.json(completely consistent with the conversation output)
  • Scoring fields (scores for each dimension, comprehensive score) retain the values calculated by the script and must not be overwritten
  • Numeric fields are filled with pure numbers (without % or units)
  • Empty value fields are left as empty string ""
• Step 7.3: Generate HTML
  • python scripts/xiaohongshu_analyzer.py generate_html
• Step 7.4: Display the HTML file content in the conversation

Multi-account comparison process

• Step 5M: Generate a diagnostic report for each account (same as step 5), and output a comparison summary (core differences, common problems, development suggestions)
• Step 6M: Display similar accounts (same as step 6)
• Step 7M: Generate multi-account HTML report
  • python scripts/xiaohongshu_analyzer.py build_multi_report_data --account_names "账号1,账号2"
  • Fill output/multi_report_data.json
  • python scripts/xiaohongshu_analyzer.py generate_multi_html
  • Display the HTML file content in the conversation

Optional branches

• User provides backend data: Directly use the data to generate the report, skipping steps 2-3
• User replies with a number to continue analyzing similar accounts: Obtain the account's Xiaohongshu ID and execute the complete process (steps 2-7)

Resource index

• Script: scripts/xiaohongshu_analyzer.py — query queries data and saves raw_data.json, build_report_data generates the report_data.json template, generate_html generates the HTML report
• Script: scripts/html_generator.py — HTML report generation logic (template replacement, field filling, conditional removal)
• Script: scripts/html_checker.py — HTML data integrity check and repair
• Reference: references/report_template.md — Must be read before generating the report, and the output must strictly follow this format
• Reference: references/api_guide.md — Read when understanding interface fields
• Reference: references/workflow_guide.md — Read when handling opening remarks, query results, similar account display, scoring rules, and data entry specifications
• Reference: references/benchmark_data.md — Read when understanding benchmark data and excellent value references for each account tier
• Asset: assets/report_template.html — Single-account HTML format report template
• Asset: assets/report_data_template.json — Single-account report data template
• Asset: assets/multi_report_template.html — Multi-account comparison HTML format report template
• Asset: assets/multi_report_data_template.json — Multi-account report data template

Notes

• The diagnostic report must be output directly in the conversation; do not generate an md file
• All dynamically generated content must be based on API-returned data + WebSearch results; fabrication is prohibited
• Output order: output the diagnostic report in the conversation → display similar accounts → generate HTML and display it
• The content of the HTML report must be completely consistent with the conversation output
• HTML report generation must be executed in step 7 and must not be omitted
• Interaction scale and update capacity are calculated based on the obtained work date difference, not a fixed 30 days
• Viral post determination is based on interaction count (likes + favorites), not just likes
• gmtCreate is the time when the Honghu data system first included it, not the account registration time