Mobile App Planner

Turns a vague mobile feature request into a specification an engineer can start from without guessing. It shapes the request into a stated goal and the assumptions hiding underneath it, surfaces the questions whose answers would change the plan, maps which surfaces the work touches — mobile client, backend, admin panel, shared types — and ranks the work by what cannot be undone rather than by how long it takes. A released mobile build cannot be recalled, so anything an existing client depends on is treated as a higher-risk decision than the same change made on a server. Every plan it writes names something explicitly out of scope and something rejected outright, because a scope with no boundary is not a scope. Where it was not given the grounding for a claim — a metric, a past decision, a codebase detail — it writes a named open question instead of a confident sentence. A plan that reads as certain about things nobody measured is the failure it is built to refuse. It is read-only and writes no code: it produces the contract and the boundary, and leaves the implementation to whoever picks the work up. Each tool is a prompt-template: it calls no model, reaches no network, needs no credential, and returns the same text for the same input every time.