Skip to content

README.md

Latest commit

 

History

History
96 lines (76 loc) · 3.16 KB

README.md

File metadata and controls

96 lines (76 loc) · 3.16 KB

Python Implementation Plugin

A Claude Code plugin providing 8 specialized agents for Python development, optimized for modern Python 3.12+ with type hints, functional patterns, and TDD methodology.

Philosophy

Deep expertise through specialized agents with clear handoffs. Each agent owns a specific domain and knows when to delegate to others.

Agent Selection Guide

Need Agent When
New feature with TDD python-coder Writing new code with tests
Integration/E2E tests python-tester Comprehensive test suites
Runtime errors python-debugger App running, errors occurring
Build/env failures python-env uv/pip/venv/Docker issues
Database/persistence python-data SQLAlchemy, Redis, migrations
Native/FFI code python-platform C extensions, subprocess
PyPI/Docker releases python-release Publishing, CI/CD, versioning
REST/GraphQL APIs python-api FastAPI, Django, endpoints

Key Design Decisions

1. python-tester Owns ALL Testing

Except inline TDD (python-coder writes unit tests with features), python-tester owns:

  • Integration tests
  • E2E tests
  • Property-based tests
  • Snapshot tests
  • Performance tests
  • Test infrastructure

2. python-env FIXES, python-release CONFIGURES

  • python-env: Broken builds, dependency conflicts, version issues
  • python-release: New release configs, CI/CD setup, publishing

3. Handoff Protocol

Each agent explicitly declares handoffs to prevent overlap.

Technology Stack

Category Tools
Package Management uv, pip, poetry
Type Checking pyright, mypy
Linting ruff, flake8, pylint
Formatting black, ruff format
Testing pytest, hypothesis, pytest-asyncio
Web Frameworks FastAPI, Django, Flask
Data SQLAlchemy, Pydantic, Redis
Functional returns, pyrsistent
CLI typer, click, rich

Agent Architecture

Python Implementation Plugin
├─ CREATE LAYER
│  └─ python-coder: New features with TDD, type hints, functional patterns
├─ TEST LAYER
│  └─ python-tester: All testing (unit, integration, e2e, property-based)
├─ DEBUG LAYER
│  └─ python-debugger: Runtime issues, profiling, logging
├─ INFRASTRUCTURE LAYER
│  ├─ python-env: Fix builds, dependencies, Docker, CI
│  └─ python-release: PyPI, Docker Hub, versioning, CHANGELOG
└─ SPECIALIZATION LAYERS
   ├─ python-data: Databases, ORMs, migrations, caching
   ├─ python-platform: Native code, FFI, multiprocessing, OS integration
   └─ python-api: REST/GraphQL APIs, authentication, rate limiting

Example Workflow

User: "Add user authentication to the FastAPI app"

1. python-coder: Implements auth with inline unit tests (TDD)
2. python-tester: Adds integration tests for auth flows
3. python-api: Configures OAuth2/JWT middleware
4. python-data: Sets up user storage, sessions
5. python-release: Updates version, CHANGELOG

Installation

/plugin install impl-python@lucid-toolkit

Usage

Agents are automatically available as subagent types in the Task tool:

Task(subagent_type="specialize:python-coder", prompt="...")