Files

SKILL.mdagentsreferences

Install

Install the containing plugin

/plugin install ci-cd@llm-skills

Invoke this skill after installation

/ci-cd:poetry

Download poetry-skill.zip

This skill is bundled inside ci-cd. Install the plugin once, then Claude Code can use any of its included skills. Browse the full plugin repository at github.com/alisonaquinas/llm-ci-dev.

SKILL.md

name: poetry description: Manage Python project dependencies, virtual environments, and package builds with Poetry. Use when tasks mention poetry commands, pyproject.toml configuration, dependency groups, or publishing packages to PyPI.

Poetry

Intent Router

Request	Reference	Load When
Install tool, first-time setup, shell completions	`references/install-and-setup.md`	Install Poetry or configure the environment
pyproject.toml fields, version constraints, extras	`references/pyproject-toml.md`	Configure project metadata or dependency specs
CLI commands, workflows	`references/command-cookbook.md`	Run add/remove/install/build/publish operations
Dependency groups, optional groups	`references/dependency-groups.md`	Define or install dev/test/docs dependency groups

Quick Start

# Create new project
poetry new my-project

# Add dependency
poetry add requests

# Install all dependencies
poetry install

# Run in virtual environment
poetry run python main.py

Core Command Tracks

New project: poetry new my-project — scaffold with pyproject.toml and package directory
Init in existing dir: poetry init — interactive pyproject.toml creation
Add dependency: poetry add requests — adds and installs, updates lock file
Add dev dependency: poetry add --group dev pytest — scoped to dev group
Install project: poetry install — installs all dependencies from lock file
Update dependencies: poetry update — resolves newest compatible versions
Build distribution: poetry build — produces wheel and sdist in dist/
Publish to PyPI: poetry publish --dry-run — validate before publishing

Safety Guardrails

Poetry creates an isolated virtual environment by default — avoid installing into the system Python.
Passing --no-root skips installing the project package itself; use only when running in CI with no editable install needed.
Commit poetry.lock to version control to ensure reproducible installs across environments.
Use poetry env use /path/to/python to pin a specific Python interpreter for a project.
Run poetry check before publishing to validate pyproject.toml consistency.

Workflow

Define project metadata and dependencies in pyproject.toml.
Run poetry install to create the virtual environment and install all packages.
Use poetry add / poetry remove to keep the lock file up to date.
Run poetry build then poetry publish --dry-run before any release.
Commit both pyproject.toml and poetry.lock.

# Troubleshoot environment issues: inspect the venv path and check dependency resolution
poetry env info
poetry check
poetry show --tree

Related Skills

pipenv — alternative workflow combining pip and virtualenv into one tool
pip — lower-level package installer used by Poetry under the hood
ci-architecture — integrating Poetry into CI/CD pipelines

References

references/install-and-setup.md
references/pyproject-toml.md
references/command-cookbook.md
references/dependency-groups.md
Official docs: https://python-poetry.org/docs/
PyPI publishing guide: https://python-poetry.org/docs/repositories/
Dependency specification: https://python-poetry.org/docs/dependency-specification/