Virtual Environment Manager Skill

Purpose

Single responsibility: Create and manage Python virtual environments for project isolation. (BP-4)

Grounding Checkpoint (Archetype 1 Mitigation)

Before executing, VERIFY:

• Python interpreter available and version known
• Target directory exists and is writable
• No conflicting venv already active
• requirements.txt or pyproject.toml exists (for install)

DO NOT create venv without confirming Python version.

Uncertainty Escalation (Archetype 2 Mitigation)

ASK USER instead of guessing when:

• Multiple Python versions available - which to use?
• Existing venv found - replace or use?
• requirements.txt vs pyproject.toml both present
• Development vs production dependencies

NEVER delete existing venv without confirmation.

Context Scope (Archetype 3 Mitigation)

Workflow Steps

Step 1: Check Python Environment (Grounding)

# Available Python versions
which python3
python3 --version

# Check for existing venv
ls -la venv/ 2>/dev/null || ls -la .venv/ 2>/dev/null || echo "No venv found"

# Current active environment
echo "VIRTUAL_ENV: $VIRTUAL_ENV"

Step 2: Create Virtual Environment

# Standard creation
python3 -m venv venv

# With specific Python version
python3.11 -m venv venv

# With system packages access
python3 -m venv venv --system-site-packages

# Verify creation
ls -la venv/bin/python

Step 3: Activate and Install Dependencies

# Activate
source venv/bin/activate

# Upgrade pip
pip install --upgrade pip

# Install from requirements.txt
pip install -r requirements.txt

# Or from pyproject.toml
pip install -e .

# Development dependencies
pip install -r requirements-dev.txt

Step 4: Validate Environment

# List installed packages
pip list

# Check for missing deps
pip check

# Export current state
pip freeze > requirements-lock.txt

Recovery Protocol (Archetype 4 Mitigation)

On error:

1. PAUSE - Don't delete corrupted venv
2. DIAGNOSE - Check error type:
   • venv creation failed → Check Python installation
   • pip install failed → Check network, package availability
   • activation failed → Check shell compatibility
   • dependency conflict → Use pip-tools or poetry
3. ADAPT - Try alternative approach (rebuild, different Python)
4. RETRY - With fresh venv (max 3 attempts)
5. ESCALATE - Report with pip debug info

Rollback command:

rm -rf venv && python3 -m venv venv

Checkpoint Support

State saved to: .aiwg/working/checkpoints/venv-manager/

checkpoints/venv-manager/
├── python_version.txt       # Python used
├── pip_freeze.txt           # Installed packages
├── creation_log.txt         # Creation output
└── validation_status.json   # Health check results

Common Commands

Best Practices

1. Naming: Use venv/ or .venv/ consistently
2. Git: Add venv to .gitignore
3. Lock files: Use pip freeze or pip-tools
4. Isolation: Always create project-specific venvs
5. Documentation: Document Python version in README

References

• Python venv docs: https://docs.python.org/3/library/venv.html
• REF-001: Production-Grade Agentic Workflows (BP-4)
• REF-002: LLM Failure Modes (Archetype 1 grounding)