Latest Version: 0.8.1

Edit Guru is a CLI tool powered by Large Language Models (LLMs) that enables intelligent file manipulation and editing through natural language commands. It acts as an AI-powered file system operator that can understand and execute complex file operations.

Example of setting up App scaffolding for a flask repo:

(source it generated is in this repo at misc/myapp)

Example usage by generating the README for this repo:

• Natural Language Task Processing: Describe your file operations in plain English
• Intelligent Planning: Automatically generates and executes plans for complex file operations
• File System Operations:
  • Text replacement across multiple files
  • File creation, editing, and deletion
  • Directory management
  • File moving and copying
  • Content search and manipulation
• Safe Operation: Built-in safeguards to prevent operations outside repository boundaries
• Interactive Mode: Review and approve operations before execution

Install Edit Guru using pipx:

pipx install git https://github.com/msull/EditGuru.git

Before using Edit Guru, set up your OpenAI API key in your environment.

Unix/Linux/MacOS:

export EDITGURU_OPENAI_API_KEY='your_api_key_here'

Windows:

set EDITGURU_OPENAI_API_KEY='your_api_key_here'

The following environment variables can be used to control the behavior of EditGuru:

• EDITGURU_OPENAI_API_KEY: (Required)Your OpenAI API key for accessing the AI functionalities.
• EDITGURU_ENABLE_BEDROCK: (Optional) Set this variable to enable the use of additional models from AWS Bedrock for the --plan-model and --model options during tool execution; must have already configured AWS Credentials with access to the desired Bedrock models.

Invoke Edit Guru using the eg command followed by your task description:

eg "your task description" [options]

• --approve: Automatically approve the generated plan and execute it
• --approve-tools: Pre-approve all tool usage during execution
• -f: Shortcut for both --approve and --approve-tools
• --plan-model: Specify a different model for plan generation
• --model: Specify the model for task execution (default: gpt-4o-mini)
• --use-cwd: Use the current working directory for file listing and operations instead of using the git repo the command is executed from

# Configuration Management: Update database settings across environments
eg "update the database connection strings in all config/*.json files to use the new hostname db-prod-v2.example.com, but only in files that don't contain 'test' in their name" --approve

# Project Setup: Initialize new feature structure with boilerplate
eg "create a new feature module named 'user_authentication' with standard files: __init__.py, models.py, views.py, and tests/test_*.py files. Add basic boilerplate code in each" -f

# File Listing: Use current directory for operations
eg "please create subdirs and organize the files in this folder" --use-cwd

• Text Replacement: Search and replace text across multiple files
• File Management: Create, read, edit, and delete files
• Directory Operations: Create and manage directory structures
• Content Manipulation: Insert, modify, or remove content at specific locations
• File Organization: Move or copy files while maintaining repository structure

• Repository boundary enforcement
• Operation preview and approval system
• Safe tool execution modes
• Error handling and validation

This project is licensed under the MIT License.