Poetry Rye Dependency Management

Poetry / Rye Dependency Management Skill

Overview

Poetry and Rye are Python dependency managers that enforce lockfile-driven, deterministic builds. Both use pyproject.toml as the single configuration file. Poetry is the established standard (since 2018); Rye is a newer Astral tool that bridges to uv. For greenfield projects, consider modern-python skill (uv-native). This skill covers Poetry/Rye for existing codebases and teams already invested in these tools.

When to Use

• When maintaining existing Poetry or Rye projects
• When a team has standardized on Poetry and migration to uv is not planned
• When publishing Python packages to PyPI (Poetry has mature publishing support)
• When managing monorepo Python workspaces
• When auditing or upgrading dependency lockfiles

Iron Laws

1. ALWAYS commit the lockfile (poetry.lock or requirements.lock) -- without it, builds are non-deterministic and CI/CD will resolve different versions than development.
2. NEVER use pip install in a Poetry/Rye-managed project -- it bypasses the resolver and creates ghost dependencies invisible to the lockfile.
3. ALWAYS use poetry add/rye add to add dependencies -- manual pyproject.toml edits without re-locking create stale lockfiles.
4. NEVER pin transitive dependencies manually -- let the solver manage the full dependency graph; pinning transitive deps causes resolver conflicts.
5. ALWAYS separate runtime and development dependencies into groups -- shipping dev/test dependencies in production images wastes space and expands attack surface.

Anti-Patterns

Workflow

Poetry Project Setup

# Initialize new project
poetry init --name my-project --python ">=3.12"

# Add dependencies by group
poetry add requests httpx
poetry add --group dev ruff pytest pytest-cov
poetry add --group docs sphinx

# Install all groups
poetry install

# Install production only
poetry install --only main

Rye Project Setup

# Initialize new project
rye init my-project
cd my-project

# Add dependencies
rye add requests httpx
rye add --dev ruff pytest pytest-cov

# Sync (install) dependencies
rye sync

pyproject.toml Configuration (Poetry)

[tool.poetry]
name = "my-project"
version = "0.1.0"
description = "Project description"
authors = ["Team <team@example.com>"]
readme = "README.md"

[tool.poetry.dependencies]
python = ">=3.12"
requests = "^2.31"
httpx = "^0.27"

[tool.poetry.group.dev.dependencies]
ruff = "^0.9"
pytest = "^8.0"
pytest-cov = "^6.0"

[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"

CI/CD Integration (GitHub Actions)

name: CI
on: [push, pull_request]
jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with:
          python-version: '3.12'
      - name: Install Poetry
        run: pipx install poetry
      - name: Cache dependencies
        uses: actions/cache@v4
        with:
          path: ~/.cache/pypoetry
          key: poetry-$HASH_OF_LOCKFILE
      - name: Install dependencies
        run: poetry install --no-root
      - name: Lint
        run: poetry run ruff check .
      - name: Test
        run: poetry run pytest --cov

Security Audit

# Poetry: audit dependencies for known CVEs
poetry audit

# Rye: use pip-audit integration
rye run pip-audit

Migration to uv

When ready to migrate from Poetry/Rye to uv:

# Export Poetry dependencies
poetry export -f requirements.txt --output requirements.txt

# Initialize uv project
uv init
uv add $(grep -v '^#' requirements.txt | grep -v '^\-' | cut -d'=' -f1)

# Verify
uv sync
uv run pytest

See modern-python skill for the complete uv workflow.

Complementary Skills

Memory Protocol (MANDATORY)

Before starting:

Read .claude/context/memory/learnings.md for prior Python packaging decisions.

After completing: Record any migration issues, version constraints, or resolver conflicts to .claude/context/memory/learnings.md.

ASSUME INTERRUPTION: Your context may reset. If it's not in memory, it didn't happen.