Jeffrey Emanuel's battle-tested approach to planning before you write a single line of code. The core idea is spending 80%+ of your time iterating on a massive markdown plan using GPT Pro Extended Reasoning, then converting it into self-contained task "beads" for Claude Code to execute. You'll run 4-5 rounds of refinement using exact prompts provided, optionally blending competing plans from Gemini, Grok, and Opus. The methodology produces 3,500+ line plans that include workflows, architecture justifications, and dependency chains. It's methodical to the point of feeling tedious, but the claim is you save 10x the time during implementation because agents have complete context without needing to ask questions.
npx skills add https://github.com/dicklesworthstone/agent_flywheel_clawdbot_skills_and_integrations --skill planning-workflow