PDFScribe
English | 繁體中文 | 简体中文
Agent skill for Claude Code that converts PDF to structured Markdown for development use.
Full pipeline — extract pages (images + text), convert to Markdown as a System Analyst, verify against source, auto-fix errors in a loop.
Features
- 4-stage pipeline: Extract → Convert → Verify → Fix-Verify Loop
- SA (System Analyst) role: output is structured for developer consumption
- Standalone Go binary for PDF extraction — no Python or runtime dependencies needed
- Cross-platform: Windows, Linux, macOS (amd64 + arm64)
- Split by chapters or single-file output
- Parallel agent processing for large documents
- Embedded image metadata with auto-classification (content vs. decorative)
- AI visual inspection of images before embedding in Markdown
- Strict fix-verify loop: re-verifies after every fix until zero issues (max 5 rounds)
Install
Add the marketplace:
/plugin marketplace add GoneTone/pdfscribe
Install the plugin:
/plugin install pdfscribe@pdfscribe
Usage
In Claude Code:
/pdf-to-md <pdf_path> [--single] [--lang <language>]
<pdf_path> — Path to PDF file (required)--single — Output single Markdown file instead of splitting by chapters--lang <language> — Output language, e.g. zh-TW, en (auto-detect by default)
Output
docs/pdfscribe/{pdf-filename}/
├── extracts/
│ ├── images/ # Page screenshots (300 DPI)
│ ├── texts/ # Extracted text per page
│ └── embedded/ # Embedded images (charts, photos)
│ └── metadata.json # Image position, size, classification
├── 00-Table-of-Contents.md
├── 01-Chapter-Name.md
└── 02-Chapter-Name.md
File names adapt to the user's language.
Build Scripts from Source
Build the PDF extraction binaries from source. Requires Go 1.21+.
# Linux / macOS / WSL / Git Bash on Windows
./build.sh
# Windows CMD
build.bat
Builds 6 binaries to skills/pdf-to-md/scripts/:
skills/pdf-to-md/scripts/
├── pdf-extract-windows-amd64.exe
├── pdf-extract-windows-arm64.exe
├── pdf-extract-linux-amd64
├── pdf-extract-linux-arm64
├── pdf-extract-darwin-amd64
└── pdf-extract-darwin-arm64
Requirements
License
MIT
GoneTone