Overview
Bilibili Comment Analysis is an intelligent comment insight tool designed specifically for content creators, UP hosts, brand operators, and researchers. Enter any Bilibili video link or BV number to retrieve comment data with one click, and quickly understand users' real feedback through four-dimensional sentiment analysis (positive/negative/demand/competitor), while generating an interactive HTML report that can be saved offline.
With this skill, you can:
- 📊 Retrieve first-level video comments page by page (20 per page, supports pagination)
- 🤖 AI automatically performs four-dimensional sentiment analysis, understanding Bilibili meme culture
- 📋 Generate a dark-themed interactive HTML report, easy to save and share
Suitable for scenarios such as UP host retrospectives, brand public opinion monitoring, competitor analysis, and content strategy research.
Features
Core Features
- Comment Retrieval: Paste a video link or BV number to pull the latest comment data
- Paginated Browsing: 20 per page, supports forward and backward pagination, browse large numbers of comments step by step
- Four-Dimensional AI Analysis: Four categories of sentiment analysis—positive/negative/demand/competitor—with representative comment quotes for each dimension
- Pinned Marking: Special marking for pinned comments, identify important content at a glance
- HTML Report: Dark-themed interactive report, supports offline saving and sharing
Highlights
- Bilibili Meme Culture Understanding: Recognize danmaku terminology, abbreviations (awsl / yyds / xswl / yeqinghui, etc.) and the real context of popular memes
- Complete Data Display: Full display of work details (views/likes/coins/favorites/danmaku, etc.) and comment tables, without truncation
- Encoding-Safe Backfill: Intelligent encoding safeguards ensure Chinese analysis content is injected into the report completely and accurately
Authentication
- Get API Key: Go to Honghu Hub
- Configuration method 1: Write to
~/.openclaw/openclaw.json→{ "env": { "REDFOX_API_KEY": "ak_xxxx..." } } - Configuration method 2: Execute in terminal
export REDFOX_API_KEY="ak_xxxx..."
Workflow
For the core execution workflow, see
references/core_workflow.md
Usage Guide
Simply state the video link or BV number in natural language; no need to memorize fixed commands.
Common Phrases Quick Reference
| Intent | Example Phrasing | Effect |
|---|---|---|
| View video comments | "View comments on Bilibili video BV1xx411c7mD" | Get page 1 comments + AI summary |
| Query via link | Paste https://www.bilibili.com/video/BV1xx411c7mD |
Automatically extract the BV number and pull comments |
| Pagination | "Next page" "Page 3" | Continue with the current video and get comments from the specified page |
| Generate report | "Needed" "Generate HTML" | Backfill AI analysis and generate an interactive HTML report |
Output Example
After the query is complete, you will receive the following in order:
- Work details(core data such as title, uploader, play count, likes, etc.)
- Comment table(all comments on the current page, including nickname links, content, like count, time, and IP location)
- Four-dimensional sentiment analysis(positive/negative/demand/competitor, each with representative comments)
- Ask whether to generate an HTML report
Use cases
| Scenario | Role | Example questions | Benefits |
|---|---|---|---|
| Content review | Uploader | "Let's see how the comments on this video are" | Quickly understand audience feedback and optimize topic selection strategy |
| Public opinion monitoring | Brand operations | "Analyze the negative comments on this video" | Identify crisis signals in time and formulate response plans |
| Competitor analysis | Content planning | "Analyze user comments on competitor videos" | Understand competitors' strengths and weaknesses and identify differentiation directions |
| User research | Product manager | "See what needs users mentioned in the comments" | Uncover real needs and guide product iteration |
Project architecture
bilibili-comment/
├── SKILL.md # 技能配置与使用说明
├── references/
│ └── core_workflow.md # Agent 核心执行流程
├── assets/
│ └── report_template.html # HTML 报告模板
└── scripts/
├── bili_comment_search.py # 评论获取脚本（调用 API，自动生成 HTML）
└── backfill_html.py # AI 分析回填脚本（将分析结果写入 HTML）
| Module | Responsibility |
|---|---|
bili_comment_search.py |
Call the RedFox API to obtain comment data and generate an HTML report containing placeholders |
backfill_html.py |
Safely backfill the AI four-dimensional analysis JSON into the HTML report, supporting direct file reading mode |
report_template.html |
Dark theme interactive report template |
Tech stack: Python standard library (urllib). No additional dependencies need to be installed.
FAQ
Installation-related
Q1: How do I obtain REDFOX_API_KEY?
A: Go to RedFox Hub to register an account and apply for an API Key, then configure it in your local environment variables as described in the authentication section.
Q2: Do I need to install additional dependencies?
A: No. The script uses the Python standard library urllib, and it can run directly in a Python 3.6+ environment.
Usage-related
Q3: How many comments can be obtained at one time?
A: Each call obtains 1 page (20 comments). If you need more data, use pagination commands to obtain it page by page, or go to RedFox Platform and contact customer service to obtain batch capabilities.
Q4: Does it support obtaining secondary comments (replies)?
A: The current version only supports first-level comments; secondary replies are not supported for now.
Troubleshooting
Q5: What should I do if some Chinese text appears garbled (question marks) in the HTML report?
A: This is a compatibility issue in the encoding backfill step. Solution: Use Python to write the analysis data as a UTF-8 JSON file, then use backfill_html.py --json-file to read and backfill directly. See details in references/core_workflow.md Step 3。
Q6: How can I get help if I have other questions?
A: Go to RedFox Platform and contact customer service, or consult the skill documentation.