Doc Mindmap - Document Intelligent Organizer 📚🧠

Batch convert scattered office documents (PDF, PPT, Word, Excel, etc.) to Markdown, generate summaries and three-dimensional classifications using local Ollama models, and present multiple classification schemes simultaneously with soft links, with zero additional disk usage.

When to Use

Use this skill when users:

Want to organize and archive a large number of documents
Need to generate summaries for a batch of documents
Want to generate mindmaps for documents
Want to convert PDF, PPT, Word to Markdown
Need document classification suggestions or directory structure solutions
Want to quickly understand what documents are in a folder
Need to detect duplicate files

Trigger Keywords: document organization, document classification, mindmap, document summary, PDF to Markdown, batch conversion, document archiving

Features

🔄 Batch Conversion - One-click conversion of PDF, PPT, Word, Excel, etc. to Markdown
📋 CSV Index - Automatically generate document index with MD5 and duplicate detection
🔍 Duplicate Detection - Identify duplicate files via MD5 comparison and suggest deletion to free up space
📝 Local Summary - Generate summaries using Ollama local models, no Claude context consumption
🗂️ Three-Dimensional Classification - Classify simultaneously by three schemes: topic/usage/client
🔗 Symlink Directory - Realize coexistence of multiple classifications via symlink with zero additional disk usage
✏️ Intelligent Rename - AI suggests clearer filenames, optimized names can be used for soft links
🛡️ Security Mechanism - Read-only conversion, no modification to original files

Supported Formats

Format	Extension	Description
📄 PDF	.pdf	PDF document
📊 PPT	.pptx	PowerPoint presentation
📝 Word	.docx	Word document
📈 Excel	.xlsx, .xls	Spreadsheet
📈 CSV	.csv	Comma-separated values
🌐 HTML	.html, .htm	Web page
📚 EPUB	.epub	E-book
📋 JSON	.json	JSON data
📋 XML	.xml	XML data

Usage

Preview Document List (Including Duplicate Detection)

bash

python scripts/doc_converter.py ~/Documents/reports --preview

Execute Conversion

bash

python scripts/doc_converter.py ~/Documents/reports --convert --confirm

Convert Specific Files

bash

python scripts/doc_converter.py file1.pdf file2.pptx --convert --confirm

Generate Summary (Requires Conversion First)

bash

python scripts/doc_converter.py ~/Documents/reports --summarize
python scripts/doc_converter.py ~/Documents/reports --summarize --model qwen3:8b

Three-Dimensional Classification + Symlink (Requires Summary First)

bash

python scripts/doc_converter.py ~/Documents/reports --organize

Classification + Use AI-Suggested Filenames

bash

python scripts/doc_converter.py ~/Documents/reports --organize --rename

Complete Full Process in One Step

bash

python scripts/doc_converter.py ~/Documents/reports --convert --confirm --summarize --organize
# With optimized filenames
python scripts/doc_converter.py ~/Documents/reports --convert --confirm --summarize --organize --rename

JSON Format Output

bash

python scripts/doc_converter.py ~/Documents --preview --json

Arguments

Argument	Description
`paths`	File or directory path (supports multiple)
`--preview`	Preview mode, list documents + duplicate detection
`--convert`	Execute batch conversion (automatically skip duplicate files)
`--summarize`	Generate summaries using Ollama local models (requires --convert first)
`--organize`	Three-dimensional classification and generate symlink directories (requires --summarize first)
`--rename`	Soft links use AI-suggested optimized filenames (works with --organize)
`--model MODEL`	Ollama model name (default: qwen2.5:3b)
`--confirm`	Confirm execution (security mechanism)
`--json`	Output in JSON format

Output Structure

Conversion output is stored in the hidden

.summaries/

directory under the source folder:

{source}/
└── .summaries/
    ├── converted/              # .md files converted by markitdown
    │   ├── report.pdf.md
    │   ├── slides.pptx.md
    │   └── data.xlsx.md
    ├── briefs/                 # Summaries generated by Ollama
    │   ├── report.pdf.brief.md
    │   ├── slides.pptx.brief.md
    │   └── data.xlsx.brief.md
    ├── schemes/                # Symlink classification directories
    │   ├── by-topic/           # Classified by topic
    │   │   ├── AI Technology/
    │   │   │   └── AI-Driven Product Management Guide.pptx -> ../../../../slides.pptx  # --rename
    │   │   └── Data Governance/
    │   │       └── C-End Data Governance Plan.pdf -> ../../../../report.pdf       # --rename
    │   ├── by-usage/           # Classified by usage
    │   │   ├── Training Materials/
    │   │   └── Client Delivery Solutions/
    │   └── by-client/          # Classified by client
    │       ├── Volvo/
    │       └── General Solutions/
    ├── mindmap.md              # Mindmap classification generated by Claude
    └── index.csv               # Conversion index (includes MD5, duplicate markers)

Dependencies

Python 3.10+
markitdown:
```
pip install 'markitdown[all]'
```

Ollama (for summary + classification):

brew install ollama

ollama pull qwen2.5:3b

requests:
```
pip install requests
```

Claude Workflow

When Claude uses this skill, follow these steps:

Step 1: Preview Documents

Run the preview command to show the document list and duplicate detection results to the user:

bash

python doc-mindmap/scripts/doc_converter.py <path> --preview

Inform the user of the number of documents found, format distribution, total size, and duplicate file status, then wait for confirmation.

Step 2: Execute Conversion

After user confirmation, execute conversion (duplicate files are automatically skipped):

bash

python doc-mindmap/scripts/doc_converter.py <path> --convert --confirm

Step 3: Generate Summary (Ollama Local Model)

Generate summaries for each document using the Ollama local model, no Claude context window consumption:

bash

python doc-mindmap/scripts/doc_converter.py <path> --summarize

It can also be executed together with convert:

bash

python doc-mindmap/scripts/doc_converter.py <path> --convert --confirm --summarize

Step 4: Three-Dimensional Classification + Symlink

Use Ollama to classify each document in three dimensions (topic/usage/client), and suggest clearer filenames for each document:

bash

# First run without --rename to show classification results and suggested filenames
python doc-mindmap/scripts/doc_converter.py <path> --organize

Show the classification results and AI-suggested filenames to the user, and ask if they want to use the optimized filenames. If the user agrees:

bash

python doc-mindmap/scripts/doc_converter.py <path> --organize --rename

Three classification schemes coexist in

.summaries/schemes/

via symlinks with zero additional disk usage.

--rename

only affects soft link names, not the original files.

Step 5: Preview Classification Results

Ask the user if they want to preview the classification directory in Finder. If the user agrees:

Copy the schemes directory to the desktop (retain symlinks):

bash

cp -a <.summaries/schemes> ~/Desktop/Document-Classification-$(date +%Y%m%d)

Open it with Finder:

bash

open ~/Desktop/Document-Classification-$(date +%Y%m%d)

Users can intuitively browse the three classification schemes in Finder, and double-click the soft link to open the original file.

Step 6: Generate Mindmap

Read the summary files in

.summaries/briefs/

and generate the

.summaries/mindmap.md

mindmap classification file.

Step 7: Present Results

Show the user:

Conversion statistics (success/failure/skipped duplicates)
List of duplicate files and deletion suggestions
Overview of three-dimensional classification
Preview of mindmap classification

Credits

markitdown - Document to Markdown tool developed by Microsoft
Ollama - Local LLM runtime framework
Use with file-organizer, classify first then organize

doc-mindmap

NPX Install

Tags

SKILL.md Content (Chinese)

Doc Mindmap - Document Intelligent Organizer 📚🧠

When to Use

Features

Supported Formats

Usage

Preview Document List (Including Duplicate Detection)

Execute Conversion

Convert Specific Files

Generate Summary (Requires Conversion First)

Three-Dimensional Classification + Symlink (Requires Summary First)

Classification + Use AI-Suggested Filenames

Complete Full Process in One Step

JSON Format Output

Arguments

Output Structure

Dependencies

Claude Workflow

Step 1: Preview Documents

Step 2: Execute Conversion

Step 3: Generate Summary (Ollama Local Model)

Step 4: Three-Dimensional Classification + Symlink

Step 5: Preview Classification Results

Step 6: Generate Mindmap

Step 7: Present Results

Credits