Jupyter Notebook Editor for AI Agents

A zero-dependency, AI-agent-friendly command-line tool for programmatically editing Jupyter Notebook (.ipynb) files. Designed specifically for Large Language Models (LLMs) and automated workflows.

Русская версия | English

---

🎯 Why This Tool?

Traditional Jupyter notebook editing requires either:

• A full Jupyter environment with heavy dependencies
• Manual JSON manipulation (error-prone and fragile)
• Complex libraries that may not work in restricted environments

This tool solves these problems by providing:

✅ Zero Dependencies - Uses only Python standard library
✅ AI-Agent Optimized - Clear CLI interface with file-based I/O pattern
✅ JSON-Safe - Preserves notebook structure and metadata
✅ Portable - Works anywhere Python 3.6+ is installed
✅ Reliable - Predictable behavior for automated workflows

---

🚀 Quick Start

Installation

No installation needed! Just download the script:

wget https://raw.githubusercontent.com/yourusername/notebook-editor/main/notebook_editor.py
chmod +x notebook_editor.py

Or clone the repository:

git clone https://github.com/yourusername/notebook-editor.git
cd notebook-editor

Basic Usage

# List all cells in a notebook
python3 notebook_editor.py list my_notebook.ipynb

# Read a specific cell with line numbers
python3 notebook_editor.py read my_notebook.ipynb 5 --numbered

# Update a cell from file
python3 notebook_editor.py update my_notebook.ipynb 5 --from-file modified_content.py

# Edit specific lines only (more efficient!)
python3 notebook_editor.py patch my_notebook.ipynb 5 --lines 10-15 --from-file patch.py

# Clear all cell outputs
python3 notebook_editor.py clear-output my_notebook.ipynb --all

# Search for content
python3 notebook_editor.py search my_notebook.ipynb "import pandas"

---

📖 Complete Command Reference

1. list - View Notebook Structure

Shows all cells with their indices, types, and content preview.

python3 notebook_editor.py list <notebook.ipynb> [--limit N] [--json]

Examples:

# Standard output
python3 notebook_editor.py list analysis.ipynb --limit 20

# JSON output (for LLM parsing)
python3 notebook_editor.py list analysis.ipynb --json

JSON output:

{
  "notebook": "analysis.ipynb",
  "total_cells": 15,
  "cells": [
    {"index": 0, "type": "code", "lines": 10, "has_output": true, ...}
  ]
}

---

2. read - Extract Cell Content

Read a specific cell's content to console or file.

python3 notebook_editor.py read <notebook.ipynb> <index> [--to-file <file>] [--numbered] [--include-output]

Examples:

# Print with line numbers (useful for patch command)
python3 notebook_editor.py read analysis.ipynb 5 --numbered

# Save to file
python3 notebook_editor.py read analysis.ipynb 5 --to-file cell_5.py

# Include execution outputs
python3 notebook_editor.py read analysis.ipynb 5 --include-output

Output with --numbered:

--- Cell 5 (code) [17 lines] ---
 1: import pandas as pd
 2: import numpy as np
 3: 
 4: def calculate_mean(data):
 5:     """Calculate the mean."""
 6:     return sum(data) / len(data)
...

---

3. search - Find Content

Search for text or regex patterns across all cells.

python3 notebook_editor.py search <notebook.ipynb> "<query>" [--regex]

Examples:

# Simple text search
python3 notebook_editor.py search analysis.ipynb "import pandas"

# Regex search
python3 notebook_editor.py search analysis.ipynb "def .*_handler" --regex

---

4. update - Modify Cell Content

Replace the entire content of a cell.

python3 notebook_editor.py update <notebook.ipynb> <index> --from-file <file>
python3 notebook_editor.py update <notebook.ipynb> <index> --content "<text>"

Examples:

# Update from file (RECOMMENDED)
python3 notebook_editor.py update analysis.ipynb 5 --from-file modified_code.py

# Keep outputs (don't clear)
python3 notebook_editor.py update analysis.ipynb 5 --from-file code.py --no-clear-output

---

5. patch - Edit Specific Lines

Replace only specified lines in a cell. Much more efficient than update!

python3 notebook_editor.py patch <notebook.ipynb> <index> --lines <range> --from-file <file>
python3 notebook_editor.py patch <notebook.ipynb> <index> --lines <range> --content "<text>"

Examples:

# Replace lines 5-10
python3 notebook_editor.py patch analysis.ipynb 3 --lines 5-10 --from-file patch.py

# Replace a single line
python3 notebook_editor.py patch analysis.ipynb 3 --lines 7-7 --content "new_value = 42"

# Insert after line 5 (add code without replacing)
python3 notebook_editor.py patch analysis.ipynb 3 --lines 5 --insert --from-file insert.py

# Disable automatic indent preservation
python3 notebook_editor.py patch analysis.ipynb 3 --lines 5-10 --from-file patch.py --no-preserve-indent

Key features:

• ✅ Automatically preserves relative indentation
• ✅ Insert mode (--insert) — adds code without replacing
• ✅ Auto-clears outputs after edit

---

6. add - Insert New Cell

Add a new code or markdown cell.

python3 notebook_editor.py add <notebook.ipynb> --type <code|markdown> --from-file <file>

Examples:

# Add at the beginning
python3 notebook_editor.py add analysis.ipynb --index 0 --type markdown --content "# Introduction"

# Add at the end (default)
python3 notebook_editor.py add analysis.ipynb --type code --from-file new_analysis.py

---

7. delete - Remove Cell

python3 notebook_editor.py delete <notebook.ipynb> <index>

---

8. diff - Preview Changes

Show what will change before updating a cell.

python3 notebook_editor.py diff <notebook.ipynb> <index> --from-file <file>

---

9. create - New Notebook

python3 notebook_editor.py create <notebook.ipynb>

---

10. clear-output - Clear Cell Outputs

Remove execution outputs from cells.

python3 notebook_editor.py clear-output <notebook.ipynb> --all
python3 notebook_editor.py clear-output <notebook.ipynb> --cells 0 2 5

Examples:

# Clear all code cells
python3 notebook_editor.py clear-output analysis.ipynb --all

# Clear specific cells
python3 notebook_editor.py clear-output analysis.ipynb --cells 0 2 5

---

11. info - Notebook Metadata

Show notebook information and statistics.

python3 notebook_editor.py info <notebook.ipynb>

Output:

Notebook: analysis.ipynb
Format: nbformat 4.5
Kernel: Python 3
Cells: 25 total
  - Code: 18
  - Markdown: 7
  - With outputs: 12
Total source lines: 450

---

12. validate - Check Structure

Validate notebook JSON structure.

python3 notebook_editor.py validate <notebook.ipynb>

Returns exit code 1 on errors — useful for CI/CD.

---

13. save-output - Extract Images

python3 notebook_editor.py save-output <notebook.ipynb> <index> --to-file <path>

---

📋 Command Summary Table

Command	Description	Key Flags
list	View structure	--limit, --json
read	Read cell	--numbered, --to-file, --include-output
search	Find text	--regex
update	Replace entire cell	--from-file, --no-clear-output
patch	Edit specific lines	--lines, --insert, --no-preserve-indent
add	Add cell	--index, --type, --from-file
delete	Remove cell	-
diff	Preview changes	--from-file
clear-output	Clear outputs	--all, --cells
info	Show metadata	-
validate	Check structure	-
create	New notebook	-
save-output	Extract images	--output-index, --to-file

---

🤖 Best Practice Workflow for AI Agents

Note for Users: There is a dedicated guide for AI agents located at README_AGENT.md.

Recommended Workflow

# 1. Explore: Understand the structure
python3 notebook_editor.py list notebook.ipynb

# 2. Read with line numbers
python3 notebook_editor.py read notebook.ipynb 5 --numbered

# 3. Precise editing (more efficient than update!)
python3 notebook_editor.py patch notebook.ipynb 5 --lines 10-15 --from-file patch.py

# OR full cell replacement
python3 notebook_editor.py read notebook.ipynb 5 --to-file temp.py
# (edit temp.py)
python3 notebook_editor.py update notebook.ipynb 5 --from-file temp.py

Advantages of patch over update

update	patch
Must copy entire cell	Edit only needed lines
Easy to break indentation	Indentation preserved automatically
Uses more tokens	Saves tokens

---

📋 Requirements

• Python 3.6+ (no external packages required)
• Works on Linux, macOS, and Windows

---

🔧 Technical Details

File Format

The tool works with standard Jupyter Notebook format (.ipynb), which is JSON-based:

• Preserves all metadata
• Maintains cell execution counts
• Handles both code and markdown cells
• Supports multi-line content with proper formatting

Safety Features

• Validates JSON structure before saving
• Creates backups implicitly (use version control!)
• Handles edge cases (empty cells, special characters, etc.)
• Provides clear error messages

---

📝 Examples

Example 1: Quick Fix for a Function

# View cell with line numbers
python3 notebook_editor.py read notebook.ipynb 3 --numbered

# Replace only lines 5-8
python3 notebook_editor.py patch notebook.ipynb 3 --lines 5-8 --content "    return x * 2"

Example 2: Add Validation to a Function

# Insert new code after line 4
echo "    if x is None:
        raise ValueError('x cannot be None')" > insert.py

python3 notebook_editor.py patch notebook.ipynb 3 --lines 4 --insert --from-file insert.py

Example 3: Pre-commit Cleanup

# Clear all outputs
python3 notebook_editor.py clear-output notebook.ipynb --all

# Validate structure
python3 notebook_editor.py validate notebook.ipynb

---

📄 License

MIT License - feel free to use in your projects!

---

Made with ❤️ for AI Agents and Developers