Skill v1.0.1
currentLLM-judged scan95/100fix(update): complete paired Claude GSD hook teardown
version: "1.0.1" name: intuitive-preflight description: Turn a vague task or plan into an execution contract with scope, non-goals, acceptance, verification, and route. disable-model-invocation: true
Intuitive Preflight
Produce an approval-ready execution contract when material scope, acceptance, verification, or route decisions remain. Reuse decisions already settled in the conversation. This skill does not implement, launch workers, create/clear a goal, or approve its own output.
Preserve a supplied plan's full intended scope. Express risk as ordering or stop gates; do not silently shrink approval to the first phase. Prefer existing surfaces and record why any new durable entity is necessary.
Ask only when missing information changes intent, scope, contracts, safety, cost, external validation, or execution permissions. Resolve mechanical, reversible details from repository evidence and state assumptions when useful.
Choose main-session execution for tiny tasks, $intuitive-refactor for selected cleanup, and durable $intuitive-flow for approved multi-stage work. Record a worker strategy only when isolation or independent work adds value.
Read verification when choosing acceptance gates; read contract when formatting the final contract. Include context, scope, non-goals, acceptance, proof, route, stop conditions, and these handoff fields: To execute:, Approval:. When blocked, replace executable handoff fields with the unresolved decisions.
After approval, follow the agreed route without rerunning preflight. Revise only sections affected by new instructions; do not request permission again for work already authorized.