knowledge-absorber
An automated research tool that cleans content, verifies facts using web search, and generates structured learning materials.
Install
mkdir -p .claude/skills/knowledge-absorber && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/6118" && unzip -o skill.zip -d .claude/skills/knowledge-absorber && rm skill.zipInstalls to .claude/skills/knowledge-absorber
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
深度解析链接、文档或代码,生成”全能导师级”的教学笔记(零基础直达精通)。具备”真理锚定”校验能力,自动识别幻觉与过时信息。支持国学风格渲染,自动清理广告和无用内容。Key capabilities
- →Ingest multi-format content
- →Verify factual accuracy
- →Generate interactive study notes
- →Apply style-based rendering
How it works
Uses a Truth Anchoring Protocol to ingest content, verify facts against web searches, and render them into structured study notes.
Inputs & outputs
When to use knowledge-absorber
- →Convert complex articles into study notes
- →Verify technical documentation accuracy
- →Summarize multi-format learning materials
- →Generate interactive HTML learning cards
About this skill
用法
把链接、文档、图片或代码整理成一份可学习、可搜索、可继续提问的学习成品,并输出:
knowledge_card.mdknowledge_card.interactive.html
只需要一条命令:
python scripts/run_full_pipeline.py “<链接或本地文件路径>” [“<更多链接或文件路径>”]
示例:
python scripts/run_full_pipeline.py “https://example.com/article”
python scripts/run_full_pipeline.py “E:\资料\AI导论.pdf”
python scripts/run_full_pipeline.py “E:\资料\架构图.png”
python scripts/run_full_pipeline.py “E:\repo\main.py”
python scripts/run_full_pipeline.py “https://example.com/article” “E:\资料\补充说明.pdf”
也支持这种自然语言写法:
python scripts/run_full_pipeline.py “学习这个 https://example.com/article”
核心流程
本技能采用 真理锚定协议 (Truth Anchoring Protocol) 确保内容准确性。
第一步:智能摄取 (Content Ingestion)
运行脚本获取干净的内容。脚本会自动清洗 HTML 噪音、广告和无用内容,并处理多模态内容(PDF/OCR)。
清理机制:
- 自动移除广告、导航、页脚、社交分享等UI元素
- 过滤包含噪音关键词的短文本(登录、注册、下载、关注等)
- 识别并移除广告容器(通过class/id模式匹配)
- 不保留明显的UI元素和营销内容
- 不保留”谢谢大佬的支持”等打赏提示
- 不保留页面导航和推荐链接
第二步:真理锚定 (Truth Anchoring)
”不要轻信任何文本,哪怕它看起来很专业。”
在生成内容之前,必须先对摄取的内容进行准确性校验。
验证流程:
- 提取核心主张:识别文中所有关键事实性主张(具体数据、API用法、历史事件、绝对化论断)
- 联网审计:调用
WebSearch验证每个主张,搜索必须包含当前年份(2026)以确保时效性 - 生成校准报告:构建”红队报告”,标注错误、过时或有争议的内容
- 显式标注:在最终输出中明确标注所有不确定或有争议的信息
验证要求:
- 每个关键主张都必须经过搜索验证
- 不得直接使用未经验证的原文内容
- 不得忽略明显的时效性问题(如”2020年最新”)
- 不得省略对绝对化论断的验证(”总是”、”从不”、”必须”)
- 不得在发现错误后不标注就直接使用
标注规范:
- 过时信息必须标注”[已过时]”
- 有争议内容必须标注”[存在争议]”
- 无法确认的内容必须标注”[待确认]”
第三步:生成教学内容
根据清洗后的内容和校准报告,生成多模态输出。
内容适配:
- 技术类:使用现代清爽风格(Apple Design 风格)
- 国学/人文类:使用水墨国学风格(传统中文字体、水墨背景、印章装饰)
国学风格触发条件: 当标题、来源、标签或内容前几段包含以下任一关键词时,自动应用国学风格:
- 经典文献:论语、庄子、道德经、史记、诗经、易经、孟子、荀子、春秋、左传、礼记、周易、尚书、大学、中庸
- 学派思想:国学、古文、佛、儒、道、哲学、人文、儒家、道家、佛家、法家、墨家、兵家、禅宗、理学、心学、玄学
- 历史人物:孔子、老子、孟子、庄子、荀子、墨子、韩非子、朱熹、王阳明、程颐、程颢
- 文学体裁:古诗、词、赋、骈文、散文、文言文
- 其他:经史子集、四书五经、诸子百家、传统文化
第四步:质量验收 (Quality Assurance)
交付前必须确认以下项目:
- ✅ HTML 包含
<script>搜索逻辑 - ✅ 国学模式下使用传统中文字体(宋体、思源宋体)和水墨风格
- ✅ 包含 Mermaid 认知地图
- ✅ 包含 5-8 个 FAQ
- ✅ 内容已清理干净(无广告、无UI元素、无打赏提示)
- ✅ 所有关键主张经过真理锚定验证
- ✅ 不确定内容有明确标注
不得出现以下情况:
- 输出包含广告或营销内容的HTML
- 输出未经验证的事实性主张
- 国学模式下使用现代西文字体
- 省略搜索功能
- 省略Mermaid图表
若任一项不符合要求,必须重新生成。
输出位置与文件
脚本会在 outputs/knowledge_YYYYMMDD_* 下生成一个输出文件夹,且最终只保留两个文件:
knowledge_card.mdknowledge_card.interactive.html
导师模式(交互 HTML 可选)
导师模式需要本地 relay 才能在浏览器里调用模型:
python scripts/mentor_relay.py(默认http://127.0.0.1:8760)
如果只是阅读/搜索正文,不需要启动 relay。
支持的输入
- URL 链接
- PDF / Word / Markdown / TXT
- PNG / JPG / JPEG / WEBP
- Python / JavaScript / TypeScript / Java / C / C++ / Go / Rust 等代码文件
- 一次传入多个链接或文件
When not to use it
- →For real-time data streams
- →When content is highly dynamic
Prerequisites
Limitations
- →Requires internet access for verification
- →Truth Anchoring may flag subjective content
How it compares
Automatically verifies facts and cleans content instead of simple summarization.
Compared to similar skills
knowledge-absorber side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| knowledge-absorber (this skill) | 1 | 4mo | Review | Intermediate |
| literature-review | 559 | 2mo | Review | Advanced |
| hugging-face-paper-publisher | 6 | 6mo | Review | Intermediate |
| youtube-summarizer | 5 | 3mo | Review | Beginner |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by YuJunZhiXue
View all by YuJunZhiXue →You might also like
literature-review
K-Dense-AI
Conduct comprehensive, systematic literature reviews using multiple academic databases (PubMed, arXiv, bioRxiv, Semantic Scholar, etc.). This skill should be used when conducting systematic literature reviews, meta-analyses, research synthesis, or comprehensive literature searches across biomedical, scientific, and technical domains. Creates professionally formatted markdown documents and PDFs with verified citations in multiple citation styles (APA, Nature, Vancouver, etc.).
hugging-face-paper-publisher
patchy631
Publish and manage research papers on Hugging Face Hub. Supports creating paper pages, linking papers to models/datasets, claiming authorship, and generating professional markdown-based research articles.
youtube-summarizer
sickn33
Extract transcripts from YouTube videos and generate comprehensive, detailed summaries using intelligent analysis frameworks
arxiv-to-md
solatis
Convert arXiv papers to LLM-consumable markdown. Invoke when user provides an arXiv ID or URL, or when syncing academic papers from a PDF folder to a markdown destination.
deep-research
richardnguyen0715
Universal deep research agent team. 13-agent pipeline for rigorous academic research on any topic. 7 modes: full research, quick brief, paper review, lit-review, fact-check, Socratic guided research dialogue, and systematic review with optional meta-analysis. Covers research question formulation, So
notebooklm
mk-knight23
Use this skill to query your Google NotebookLM notebooks directly from Claude Code for source-grounded, citation-backed answers from Gemini. Browser automation, library management, persistent auth. Drastically reduced hallucinations through document-only responses.