tumee-codeguard 0.2.3

A code validation tool for enforcing access control on code modifications

TuMee CodeGuard

A powerful file change detection tool that identifies, tracks, and validates code modifications with a focus on respecting designated "guarded" regions across multiple programming languages.

Package Name: tumee-codeguard Command: codeguard

Installation

Prerequisites

• Python 3.10 or higher
• Git (for version control integration)

Setup

On macOS/Linux

# Make the install script executable
chmod +x install.sh

# Run the installation script
./install.sh

On Windows

# Run the installation script
install.bat

Usage

Running CodeGuard

On macOS/Linux

# Run CodeGuard
./run_codeguard.sh [command] [options]

# For example, to get help
./run_codeguard.sh --help

# To get effective permissions for a file
./run_codeguard.sh acl /path/to/file

On Windows

# Run CodeGuard
run_codeguard.bat [command] [options]

# For example, to get help
run_codeguard.bat --help

# To get effective permissions for a file
run_codeguard.bat acl C:\path\to\file

Commands

• acl - Get effective permissions for a file or directory
• batch-acl - Get permissions for multiple paths in a batch
• aiattributes - Manage directory-level guard annotations via .ai-attributes files
• list-guarded-directories - List directories with guard annotations
• verify - Compare two files directly
• verify-disk - Compare modified file against current version on disk
• verify-git - Compare against last checked-in version in git
• verify-revision - Compare against specific revision
• scan - Batch operations on directories
• install-hook - Install git pre-commit hook
• serve - Start MCP server

Examples

Managing Directory-Level Guard Annotations

# Create or update an .ai-attributes file
./run_codeguard.sh aiattributes create --directory ./src --rule "*.py:AI-RO" --description "*.py:Python files are AI read-only"

# List rules from .ai-attributes files
./run_codeguard.sh aiattributes list --directory ./src --recursive

# Validate .ai-attributes files
./run_codeguard.sh aiattributes validate --directory ./src --recursive

Getting Access Control Information

# Get effective permissions for a file
./run_codeguard.sh acl ./src/main.py --verbose

# Get permissions for multiple paths in a batch
./run_codeguard.sh batch-acl ./src/main.py ./src/utils.py ./src/config.py

# List directories with guard annotations
./run_codeguard.sh list-guarded-directories --directory ./src

Code Validation

# Compare two files directly
./run_codeguard.sh verify --original ./original.py --modified ./modified.py

# Compare against the last checked-in version in git
./run_codeguard.sh verify-git --file ./src/main.py

# Compare against a specific revision
./run_codeguard.sh verify-revision --file ./src/main.py --from-revision HEAD~3

Guard Annotation System

CodeGuard supports a standardized guard notation that works across programming languages:

@GUARD:WHO-PERMISSION

Where:

• @GUARD: The prefix that identifies this as a guard directive
• WHO: Indicates who the rule applies to (AI for AI systems, HU for human developers, ALL for both)
• PERMISSION: Specifies the permission level (RO for read-only, ED for editable with reason, FX for fixed/unchangeable)

Examples

# @GUARD:AI-RO This is an AI read-only section
def sensitive_function():
    pass

# @GUARD:HU-ED Humans can edit this section with good reason
def editable_function():
    pass

"""
@GUARD:ALL-FX
This section is fixed for everyone
"""

Directory-Level Guard System

CodeGuard supports directory-level guard annotations through .ai-attributes files:

# All files in this directory are AI read-only
* @GUARD:AI-RO

# All Python files in this directory and subdirectories are fixed
**/*.py @GUARD:ALL-FX

# Test files in the tests directory can be edited
tests/* @GUARD:ALL-ED