Lire les annotations PDF

---

name: read-annotated-pdf description: Extract and display annotations (highlights, comments, sticky notes, strikeouts) from a PDF file. Use when the user asks to read annotations, check PDF comments, review highlights, or says "看看我的批注". argument-hint: <path-to-pdf> allowed-tools: Bash(python3 *)

Extract annotations from the given PDF using the bundled script.

Instructions

1. Run the extraction script on the provided PDF path:

   python3 ${CLAUDE_SKILL_DIR}/scripts/read_annotations.py $ARGUMENTS
   

2. If pymupdf is not installed, install it first:

   pip install pymupdf
   

3. Parse the output. Each annotation has:

   • page: page number
   • type: Highlight, Text (sticky note), FreeText, StrikeOut, Underline, Squiggly
   • highlighted_text: the text that was highlighted/marked
   • comment: the user's written comment
   • context_before / context_after: surrounding text for locating the annotation

4. Group annotations by page and present them clearly to the user.

5. For each annotation with a comment, treat it as a revision instruction:

   • "rewrite" or "改" → rewrite that section
   • "delete" or "删" → remove that content
   • A question → answer it and propose an edit
   • Just a highlight with no comment → flag it for the user to clarify

6. After presenting all annotations, ask: "Should I apply these changes?" / "要我按这些批注改吗？"

Important

• Never recompile a PDF over the annotated file — annotations are stored in the PDF and will be lost. Always compile to a new filename or versioned name.
• When applying changes, work on the .tex source, not the PDF.