What problem does it solve? Breaking a large story or task into parallelizable implementation steps is error-prone when done ad hoc, and plans often miss dependencies or unresolved questions. This Skill delegates decomposition to a planner subagent and refines the result through an interactive feedback loop before orchestration begins. ## Core Features & Use Cases - Planner delegation: Invokes a planner subagent to analyze the codebase and produce a machine-readable orchestration-plan.json plus timestamped Markdown snapshots. - Interactive refinement loop: Presents steps, dependency graphs, risks, and open questions, then re-invokes the planner with user feedback until approval. - Ticket-level artifacts: Stores the mutable plan JSON at the ticket level so it can be shared across later orchestration runs. - Use Case: Given a feature story like "add OAuth login", produce a step-by-step plan showing which steps run in parallel, which are sequential, and what questions remain, then hand it off to orchestrate-dev. ## Quick Start Ask the agent to break your story or task description into independently orchestrable implementation steps using the plan-orchestrable-steps skill.