Development Workflow Skill Guide

Description

The dev-workflow skill ensures consistent project setup, version control, and adherence to organizational standards. Covering everything from Git branching to AI project scaffolding, these workflows are mandatory for all new projects and collaboration tasks.

When to Use the Skill

• •Setting up a repository or project: Ensure correct structures, environments, and configurations.
• •Version control workflows: Commit, branch management, and Gitflow adherence.
• •AI/ML-specific needs: Proper Python package management, environments, and coding standards.
• •Synchronizing or updating project artifacts: Ensure clean execution of project syncs.

Usage Guide

Setting Up AI/ML Projects

Scaffold a New Project

uv init --lib <project_name>
uv python pin 3.12

• •Creates the project structure and pins the Python version.
• •Populate pyproject.toml and configure uv.lock for dependencies.

Example Gitignore

# Python artifacts
__pycache__/
.venv
*.pyc

# Data directories
data/

Git Workflow

Branch Creation

Keep branch names descriptive of the feature or fix:

git checkout -b feat/user-auth
git checkout -b fix/login-timeout

• •Feature/fix branches merge to develop.
• •Ensure all commits follow the Conventional Commits format.

Commit Workflow

1. •Use git status and git diff to identify changes.
2. •Avoid staging sensitive files (e.g., .env, credentials).
3. •Write a semantic commit message (example):
   bash

   git commit -m "fix(login): handle timeout edge cases"
   

Inputs and Outputs

Inputs

• •Repository context: Current Git state (branches, changes).
• •Project environment: Python version, package manager configurations, or AI frameworks required.

Outputs

• •Git changes: Staged/stable commits reflecting best practices.
• •Project files: Well-organized structure based on uv or repo templates.

Best Practices and Version History

Best Practices

• •Git: Always work on feature branches; avoid direct main commits.
• •AI Projects: Use uv for reproducible environments; pin Python versions.

Version History