planning-workflow

<!-- TOC: Philosophy | THE EXACT PROMPT | Process Overview | References --> # Planning Workflow — The Foundation of Agentic Development > **Core Philosophy:** "Planning tokens are a lot fewer and cheaper than implementation tokens." > > The models are far smarter when reasoning about a detailed plan that fits within their context window. This is the key insight behind spending 80%+ of time on planning. --- ## Outcome — When This Skill Has Delivered You have a usable plan when **all** of the following hold: - The plan is **self-contained**: a fresh agent who has never seen the project can read it and start implementing without asking the human for clarification. - It is **dependency-aware**: every non-trivial task names what blocks it and what it unblocks, so the work decomposes cleanly into a beads graph. - It is **justified**: every architectural choice and every non-obvious feature has at least one sentence on *why*, not just *what* — future agents need the rationale to make consistent local decisions. - It has survived **at least 4 review rounds** by a strong reasoning model (GPT Pro Extended Reasoning is the proven choice) and reached steady-state — i.e., the most recent round produces marginal rather than structural revisions. - It has been **converted to beads** with the dependency graph intact, so implementation agents can pick up ready work via `br ready --json` without re-reading the plan. You have NOT delivered if any of these is true: the plan is < ~1,500 li...