What problem does it solve?
Planning turns ambiguous or unlocked decisions into a clear, reviewable execution plan and prepares only the next phase for deterministic validation and execution, preventing agents from skipping approvals or creating unauditable work.
Core Features & Use Cases
- Discovery and Synthesis: Reads CONTEXT.md and institutional learnings, maps the codebase, and writes discovery.md and approach.md to close the gap between requirements and implementation.
- Phase Planning & Approval Gate: Produces a phase-plan.md that describes 2–4 meaningful phases, prompts for human approval, and enforces a hard gate before creating execution artifacts.
- Current-Phase Preparation: After approval, writes the current-phase contract, story map, and uses the br CLI to create normalized beads with verification metadata for the approved phase only.
- Tool Integration: Integrates optional gkg discovery and recommends running br and bv where available to create and validate bead graphs.
Quick Start
Read history/<feature>/CONTEXT.md, surface institutional learnings, produce history/<feature>/phase-plan.md, get user approval, and then prepare the approved phase's contract, story map, and beads.