Image Breaker
Convert documents, PDFs, images, and web content into structured markdown notes saved to workspace and synced to Obsidian.
Workflow
1. Extract Content
For URLs/PDFs:
CODEBLOCK0
For images:
CODEBLOCK1
For already-analyzed content:
CODEBLOCK2
2. Structure the Content
Convert raw content into organized markdown:
Sections to create:
- - Overview - What is this document/content about?
- Key Points - Bullet list of main takeaways
- Detailed Breakdown - Organized subsections with headers
- Reference Ranges/Standards (if applicable) - Tables for numerical data
- Action Items (if applicable) - What to do with this information
- Source - Original URL or document name
Formatting guidelines:
- - Use tables for numerical data (reference ranges, standards, comparisons)
- Use bullet lists for key points
- Use headers (##, ###) for organization
- Include code blocks for technical content
- Bold important terms on first mention
3. Save and Sync
Create the markdown note with proper frontmatter and save to workspace:
CODEBLOCK3
Tag Assignment
Auto-assign 3 most relevant tags based on content:
Common tags:
- -
research - Academic papers, studies, references - INLINECODE1 - Lab results, biomarkers, panels
- INLINECODE2 - NMR lipid panels specifically
- INLINECODE3 - Cholesterol and lipid-related
- INLINECODE4 - BPC-157, TB-500, etc.
- INLINECODE5 - Vitamins, minerals, compounds
- INLINECODE6 - Treatment/optimization protocols
- INLINECODE7 - Business/entrepreneur health content
- INLINECODE8 - Anti-aging, healthspan
- INLINECODE9 - Cognitive/physical optimization
- INLINECODE10 - Exercise, workouts
- INLINECODE11 - Nattokinase, Toku Flow related
Prioritize specific tags over generic ones.
Output Directories
Default: INLINECODE12
Content-specific alternatives:
- - Research documents →
research/papers/ or INLINECODE14 - Lab results → INLINECODE15
- Marketing materials → INLINECODE16
- Training content → INLINECODE17
- Business documents → INLINECODE18
Choose the most appropriate directory based on content type.
Example Usage
User provides Labcorp NMR document URL:
- 1. Extract content using INLINECODE19
- Structure into markdown with:
- Overview of what NMR measures
- Key reference ranges table
- Interpretation guide
- Comparison to standard lipids
- 3. Assign tags:
bloodwork, nmr, INLINECODE22 - Save to INLINECODE23
- Sync to Obsidian vault at INLINECODE24
- Report to user with file path and Obsidian link
Best Practices
- - Always extract content first - Use web_fetch or image tool before structuring
- Create comprehensive notes - Include context, not just raw data
- Use tables for data - Reference ranges, comparisons, standards
- Tag intelligently - Maximum 3 tags, most specific/relevant
- Choose output directory wisely - Match content type to workspace organization
- Auto-sync by default - User wants notes in Obsidian for cross-referencing
- Report file location - Give user both workspace and Obsidian paths
Output Message Template
After completing the workflow:
CODEBLOCK4
Integration with Other Skills
Obsidian Sync: Automatically called after note creation
Paper Fetcher: If user provides DOI, use paper-fetcher first, then break down the PDF
Research Automation: Can batch-process multiple documents from research runs
图像分解器
将文档、PDF、图像和网页内容转换为结构化的Markdown笔记,保存到工作区并同步至Obsidian。
工作流程
1. 提取内容
对于URL/PDF:
使用web_fetch提取内容
对于图像:
使用图像工具分析和提取文本
对于已分析的内容:
用户可能直接粘贴内容,或您已提取完毕
2. 结构化内容
将原始内容转换为组织有序的Markdown:
需创建的章节:
- - 概述 - 此文档/内容是关于什么的?
- 关键要点 - 主要收获的要点列表
- 详细分解 - 带标题的组织化子章节
- 参考范围/标准(如适用)- 数值数据的表格
- 行动项(如适用)- 如何处理此信息
- 来源 - 原始URL或文档名称
格式指南:
- - 使用表格呈现数值数据(参考范围、标准、对比)
- 使用要点列表呈现关键点
- 使用标题(##、###)进行组织
- 技术内容包含代码块
- 首次提及重要术语时加粗
3. 保存与同步
创建带有正确前置元数据的Markdown笔记并保存到工作区:
python
准备前置元数据
date = 2026-02-10
tags = [research, bloodwork, nmr] # 基于内容自动分配
title = NMR脂质面板参考范围
构建完整Markdown内容
content = f---
date: {date}
tags:
- {tag1}
- {tag2}
- {tag3}
source: {original
urlor_source}
type: image-breaker-note
{title}
概述
[此文档内容的简要描述]
关键要点
[主要章节]
[带子章节的详细内容]
参考
- - 来源: [URL或文档名称]
- 提取日期: {date}
保存到工作区
output_dir = research/image-breaker-notes # 默认
或用户指定:research/bloodwork, content/references等
写入文件
filepath = f{output_dir}/{date}-{slugified-title}.md
write(filepath, content)
同步至Obsidian(使用obsidian-sync技能)
exec: python3 skills/obsidian-sync/scripts/sync
toobsidian.py {filepath} /Users/biohacker/Desktop/Connections ImageBreaker
标签分配
基于内容自动分配3个最相关的标签:
常用标签:
- - research - 学术论文、研究、参考文献
- bloodwork - 化验结果、生物标志物、面板
- nmr - 特指NMR脂质面板
- cholesterol - 胆固醇和脂质相关
- peptides - BPC-157、TB-500等
- supplements - 维生素、矿物质、化合物
- protocols - 治疗/优化方案
- founders - 商业/创业健康内容
- longevity - 抗衰老、健康寿命
- performance - 认知/身体优化
- training - 运动、锻炼
- toku - 纳豆激酶、Toku Flow相关
优先使用具体标签而非通用标签。
输出目录
默认: research/image-breaker-notes/
内容特定替代目录:
- - 研究文档 → research/papers/ 或 research/protocols/
- 化验结果 → research/bloodwork/
- 营销材料 → content/references/
- 训练内容 → research/training/
- 商业文档 → projects/business-docs/
根据内容类型选择最合适的目录。
使用示例
用户提供Labcorp NMR文档URL:
- 1. 使用web_fetch提取内容
- 结构化Markdown,包含:
- NMR测量内容的概述
- 关键参考范围表格
- 解读指南
- 与标准脂质的对比
- 3. 分配标签:bloodwork、nmr、research
- 保存至research/image-breaker-notes/2026-02-10-nmr-lipid-panel-reference.md
- 同步至Obsidian库ImageBreaker/2026-02-10-nmr-lipid-panel-reference.md
- 向用户报告文件路径和Obsidian链接
最佳实践
- - 始终先提取内容 - 在结构化之前使用web_fetch或图像工具
- 创建全面笔记 - 包含上下文,而不仅仅是原始数据
- 数据使用表格 - 参考范围、对比、标准
- 智能分配标签 - 最多3个标签,最具体/相关
- 明智选择输出目录 - 将内容类型与工作区组织匹配
- 默认自动同步 - 用户希望在Obsidian中交叉引用笔记
- 报告文件位置 - 同时提供工作区和Obsidian路径
输出消息模板
完成工作流程后:
✅ 文档已分解并保存
📝 标题: [笔记标题]
📂 位置: research/image-breaker-notes/2026-02-10-note-title.md
🔗 Obsidian: ImageBreaker/2026-02-10-note-title.md
🏷️ 标签: tag1, tag2, tag3
已创建章节:
该笔记现已在您的Obsidian库中,可进行标签和交叉引用。
与其他技能的集成
Obsidian同步: 笔记创建后自动调用
论文获取器: 如果用户提供DOI,先使用paper-fetcher,再分解PDF
研究自动化: 可批量处理来自研究运行的多个文档