python-package-runtime

# AI Builder - Python Package Runtime Use this skill to install Python packages and run Python-based helpers in a consistent repo-safe way. ## When to Use This Skill - A task needs a new Python package installed in this repo - A helper script should be run with the repo Python environment - A Python package exposes a CLI that should be run from the engine virtualenv ## Your Roles in This Skill - **Platform Engineer**: Keep Python package installation and runtime behavior consistent with the repo environment - **Backend Developer**: Run scripts and package CLIs with the correct interpreter and virtualenv paths - **Technical Writer**: Document the exact command conventions clearly and tersely ## Role Communication As an expert in your assigned roles, you must announce your actions before performing them using the following format: As a {Role, and Role-XYZ if have more roles}, I will {action description} This communication pattern ensures transparency and allows for human-in-the-loop oversight at key decision points. ## Instructions Follow these steps in order: ### Step 1: Decide what kind of Python action is needed Choose one path: 1. Install one or more new Python packages 2. Run a Python helper script 3. Run a package-provided CLI from the engine virtualenv ### Step 2: Install packages with the repo wrapper only For any new Python package in this repo, always install it from repo root with: ```bash core/bin/uv-install <package> [package...] ``` This wrapper runs `uv add` in `core/engine/`. Do not use `pip install` directly for new packages. ### Step 3: Run Python scripts with the repo interpreter Run helper scripts with: ```bash core/bin/python path/to/script.py ``` `core/bin/python` points to `core/engine/.venv/bin/python`, so scripts run with the repo-managed environment and installed packages. ### Step 4: Run package CLIs from the engine virtualenv When a package provides a CLI entrypoint, run it from: ```bash core/engine/.venv/bin/<cli-name> ``` If you need to confirm the binary name, inspect `core/engine/.venv/bin/` first. ### Step 5: Keep helpers and temp artifacts contained - Put temporary helper scripts and intermediate files under `.skillpilot/temp/` - Keep commands relative to repo root unless the task requires another working directory - Prefer small, one-purpose helpers over ad hoc command variations ## Expected Output - The required package is installed through `core/bin/uv-install`, if installation was needed - Python scripts run through `core/bin/python` - Package CLIs run from `core/engine/.venv/bin/` - The result clearly states which path was used ## Key Principles - Use one installation path: `core/bin/uv-install` - Use one Python interpreter path: `core/bin/python` - Use one CLI path convention: `core/engine/.venv/bin/<cli-name>` - Do not bypass the repo environment with direct `pip install` or a random system Python