vexy-overnight 1.1.2

No project description provided

Vexy Overnight Manager (vomgr)

A unified management tool for AI assistant CLIs (Claude, Codex, Gemini), providing streamlined launching, automated continuation workflows, and configuration synchronization.

Overview

Vexy Overnight Manager (vomgr) consolidates and simplifies the management of multiple AI assistant CLIs, replacing a collection of over-engineered legacy tools with a single, maintainable Python package. It handles:

• Unified Launching: Start Claude, Codex, or Gemini with consistent interfaces
• Automated Continuation: Smart session continuation when tasks complete
• Configuration Management: Safe editing of CLI configuration files
• Rules Synchronization: Keep instruction files (CLAUDE.md, AGENTS.md, etc.) in sync
• Tool Updates: Manage updates for all CLI tools from one place

Features

Core Commands

• vomgr install - Set up continuation hooks and configurations
• vomgr enable/disable <tool> - Toggle continuation automation
• vomgr run <tool> - Launch AI assistants with proper settings
• vomgr rules - Synchronize instruction files across projects
• vomgr update - Update CLI tools and the package itself
• vomgr status - View current configuration state
• version-bump - Automated semantic version tagging for Git repositories

Simplified Launchers

• vocl - Launch Claude with optimized settings
• voco - Launch Codex with profile management
• voge - Launch Gemini with appropriate flags

Continuation Tools

• vocl-go - Auto-continue Claude sessions (replaces 1500+ line claude4ever.py)
• voco-go - Auto-continue Codex sessions (replaces complex codex4ever.py)
• voge-go - Gemini continuation (when API available)

Installation

# Install from PyPI
pip install vexy-overnight

# Or with uv (recommended)
uv add vexy-overnight

# Install and configure
vomgr install

Quick Start

# Enable continuation for Claude
vomgr enable claude

# Launch Claude with continuation enabled
vocl
# Or
vomgr run claude

# Sync instruction files in current project
vomgr rules sync

# Update all CLI tools
vomgr update --cli

# Check status
vomgr status

Usage Examples

Managing Continuation Hooks

# Enable auto-continuation for Claude and Codex
vomgr enable claude
vomgr enable codex

# Disable continuation for specific tool
vomgr disable claude

# Check what's enabled
vomgr status

Instruction File Management

# Sync instruction files (CLAUDE.md, AGENTS.md, etc.) in current directory
vomgr rules sync

# Append text to all instruction files
vomgr rules append "Additional instructions here"

# Search in instruction files
vomgr rules search "pattern"

# Replace text across instruction files
vomgr rules replace "old text" "new text"

# Manage global instruction files in home directory
vomgr rules --global sync

Version Management

# Simple version bump (patch increment)
version-bump

# Verbose output showing all operations
version-bump --verbose

# Example workflow:
# 1. Make changes to your code
# 2. Commit changes: git add . && git commit -m "Add new feature"
# 3. Bump version: version-bump
# 4. Version tag is created and pushed automatically

The version-bump tool automatically:

• Detects the latest version tag (e.g., v1.2.3)
• Increments the patch version (e.g., v1.2.4)
• Creates a new Git tag
• Pushes the tag to the remote repository

Requirements:

• Clean working directory (no uncommitted changes)
• Git repository with remote configured
• Existing version tags in vX.Y.Z format (or starts with v1.0.0)

Launching AI Assistants

# Direct launchers (installed as console scripts)
vocl                    # Launch Claude
voco -m gpt5           # Launch Codex with gpt5 profile
voge                    # Launch Gemini

# Via vomgr
vomgr run claude --cwd /path/to/project
vomgr run codex --profile o3
vomgr run gemini

Updates and Maintenance

# Check for updates
vomgr update --check

# Update CLI tools (claude, codex, gemini)
vomgr update --cli

# Update vexy-overnight itself
vomgr update --self

# Update everything
vomgr update --all

# Dry run (show what would be updated)
vomgr update --cli --dry-run

Architecture

Simplified Design

Unlike the legacy tools with 1500+ lines of complex async code, vexy-overnight:

• No iTerm2 dependency: Uses standard subprocess calls
• No TTS: Simple logging instead of speech synthesis
• No state machines: Straightforward procedural flow
• Minimal dependencies: Just essential packages
• Testable: Every component is unit-testable
• Maintainable: Clear, simple code under 200 lines per file

Configuration Safety

• Creates backups before any config modification
• Validates changes after editing
• Provides rollback on errors
• Preserves all existing user settings
• Uses proper JSON/TOML libraries (no regex hacks)

Migration from Legacy Tools

If you're using the old tools (claude4ever.py, codex4ever.py, etc.):

# Back up existing configurations
vomgr install --backup-legacy

# Migration automatically preserves your settings
vomgr install --migrate

# Old tools remain available until you're ready
# Both can coexist during transition

Development

This project uses modern Python packaging with uv:

# Clone repository
git clone https://github.com/vexyart/vexy-overnight
cd vexy-overnight

# Set up development environment
uv venv --python 3.12
uv sync

# Run tests
python -m pytest -xvs

# Run with coverage
python -m pytest --cov=src --cov-report=term-missing

# Type checking
uvx mypy src/

# Format code
uvx ruff format src/ tests/

Requirements

• Python 3.12+
• One or more AI CLI tools installed:
  • Claude Code (npm install -g @anthropic-ai/claude-code)
  • Codex (brew install codex or from source)
  • Gemini CLI (npm install -g @google/gemini-cli)

License

MIT License

Contributing

Contributions welcome! Please ensure:

• All tests pass
• 80%+ code coverage
• Type hints on all functions
• No functions over 20 lines
• No files over 200 lines