Image Breaker

Extract and break down content from web documents, PDFs, images, and URLs into structured markdown notes stored locally and synced to Obsidian. Use when the...

Security Scan

The skill's code and instructions are consistent with its stated purpose (extract content, create markdown notes, and sync to Obsidian); it requests no unrelated credentials and has no install steps, though it assumes a local Obsidian vault path and calls an external obsidian-sync script which you should verify before use.

Purpose & Capability

Name/description (extract content and save/sync notes) matches the included instructions and helper script: it uses web_fetch/image tools to extract content, converts to markdown, saves to workspace, and invokes an obsidian-sync script to sync. No unrelated credentials, binaries, or install steps are requested.

Instruction Scope

SKILL.md instructs the agent to extract content (web_fetch, image tool), produce structured markdown, save files locally, and call an obsidian-sync script. This is within scope, but examples and the Python helper include a hardcoded local Obsidian vault path (/Users/biohacker/Desktop/Connections) and an exec call to skills/obsidian-sync/scripts/sync_to_obsidian.py; these assumptions may not match every user and could fail or attempt to access a path that doesn't exist. The skill does not instruct reading arbitrary unrelated files or exfiltrating data to external endpoints.

Install Mechanism

No install spec (instruction-only plus a small helper script). No downloads or archive extraction. The included Python script is local and simple; it does not fetch or execute remote code.

Credentials

No environment variables or credentials are required. The script uses a workspace-derived path and a default output directory; it does not request tokens/keys or other unrelated secrets.

Persistence & Privilege

The skill is user-invocable and not always-enabled. It does not request elevated persistence, nor does it modify other skills' configurations. It does call another skill's script (obsidian-sync) if present, but that is external to this skill.

Assessment

This skill appears to do what it says: extract content, create markdown notes, and optionally sync them to Obsidian. Before installing or running it: 1) Verify or change the Obsidian vault path (the example and the script hardcode /Users/biohacker/Desktop/Connections) to a path you control; otherwise the sync will fail or target an unexpected location. 2) Confirm the obsidian-sync script referred to (skills/obsidian-sync/scripts/sync_to_obsidian.py) is present and trustworthy — the skill will call that script to sync files, so that script determines what network or filesystem actions actually occur. 3) Be aware the templates include medical/protocol content (dosing, lab interpretation); the skill can produce notes that look like clinical guidance — treat such outputs cautiously and do not rely on them for medical decisions. 4) No secrets are requested by this skill, and it does not appear to exfiltrate data to remote endpoints itself, but web_fetch/image extraction will fetch remote content you provide; ensure you only process documents you are comfortable sharing with the agent. If you want a safer setup, change the default output-dir and obsidian_vault to known locations and review the obsidian-sync implementation before enabling automatic sync.

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:

Use web_fetch to extract content

For images:

Use image tool to analyze and extract text

For already-analyzed content:

User may paste content directly or you've already extracted it

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:

# Prepare frontmatter
date = "2026-02-10"
tags = ["research", "bloodwork", "nmr"]  # Auto-assigned based on content
title = "NMR Lipid Panel Reference Ranges"

# Build full markdown content
content = f"""---
date: {date}
tags:
  - {tag1}
  - {tag2}
  - {tag3}
source: {original_url_or_source}
type: image-breaker-note
---

# {title}

## Overview
[Brief description of what this document is]

## Key Points
- Point 1
- Point 2
- Point 3

## [Main Section]
[Detailed content with subsections]

## Reference
- **Source:** [URL or document name]
- **Extracted:** {date}
"""

# Save to workspace
output_dir = "research/image-breaker-notes"  # Default
# or user-specified: "research/bloodwork", "content/references", etc.

# Write file
filepath = f"{output_dir}/{date}-{slugified-title}.md"
write(filepath, content)

# Sync to Obsidian (using obsidian-sync skill)
exec: python3 skills/obsidian-sync/scripts/sync_to_obsidian.py {filepath} /Users/biohacker/Desktop/Connections ImageBreaker

Tag Assignment

Auto-assign 3 most relevant tags based on content:

Common tags:

• research - Academic papers, studies, references
• bloodwork - Lab results, biomarkers, panels
• nmr - NMR lipid panels specifically
• cholesterol - Cholesterol and lipid-related
• peptides - BPC-157, TB-500, etc.
• supplements - Vitamins, minerals, compounds
• protocols - Treatment/optimization protocols
• founders - Business/entrepreneur health content
• longevity - Anti-aging, healthspan
• performance - Cognitive/physical optimization
• training - Exercise, workouts
• toku - Nattokinase, Toku Flow related

Prioritize specific tags over generic ones.

Output Directories

Default: research/image-breaker-notes/

Content-specific alternatives:

• Research documents → research/papers/ or research/protocols/
• Lab results → research/bloodwork/
• Marketing materials → content/references/
• Training content → research/training/
• Business documents → projects/business-docs/

Choose the most appropriate directory based on content type.

Example Usage

User provides Labcorp NMR document URL:

1. Extract content using web_fetch
2. Structure into markdown with:
   • Overview of what NMR measures
   • Key reference ranges table
   • Interpretation guide
   • Comparison to standard lipids
3. Assign tags: bloodwork, nmr, research
4. Save to research/image-breaker-notes/2026-02-10-nmr-lipid-panel-reference.md
5. Sync to Obsidian vault at ImageBreaker/2026-02-10-nmr-lipid-panel-reference.md
6. 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:

✅ **Document broken down and saved**

📝 **Title:** [Note Title]
📂 **Location:** research/image-breaker-notes/2026-02-10-note-title.md
🔗 **Obsidian:** ImageBreaker/2026-02-10-note-title.md
🏷️  **Tags:** tag1, tag2, tag3

**Sections created:**
- Overview
- Key Points  
- [Main sections listed]
- Reference

The note is now in your Obsidian vault for tagging and cross-referencing.

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