The write-plan skill provides a standardized framework for Claude to follow before executing any code changes. It mandates a multi-step process: assessing component readiness, identifying knowledge gaps through a 'confidence gate', and drafting detailed plans using specialized templates for project structure, tech stack, and architecture. By requiring explicit testing strategies—including unit, integration, and E2E tests—and prioritizing regression-first bug fixes, it ensures that AI-generated work is grounded in evidence, documented thoroughly, and verified against specific requirements.
Key Features
01Mandatory confidence-gate assessment to eliminate AI guessing
02Comprehensive testing strategy enforcement including regression-first bug fixes
03Hypothesis tracking and human-verification gate identification
04Automated task dependency mapping and verification specifications
059 GitHub stars
06Standardized templates for visualizing project structure and architecture
