Relentless plan review, one question at a time
Grill Me Skill turns vague plans into explicit decisions.
Use a focused AI interviewer to examine a plan, design, implementation path or product idea. It asks targeted questions, follows branches, recommends practical answers, and keeps going until weak assumptions are visible.
- Style
- One question
- Depth
- Branching
- Output
- Better plans
Plan pressure test
Question, answer, branch, recommend, then continue until the plan holds up.
What it is
A structured interview mode for plans that are not ready yet.
Grill Me Skill is designed for moments when a user has an idea, proposal, technical approach or feature design but has not fully explored the choices hidden inside it. The skill changes the assistant's job from "help me do it" to "interrogate whether this plan is complete enough to do well."
The referenced skill file emphasizes pressure-testing. The assistant should ask a single question, wait for the answer, branch from that answer, explore the tree of consequences, and recommend what it would choose when a decision point has a likely best path.
Relentless questioning
The skill keeps asking until priorities, constraints, risks and tradeoffs are explicit.
Branch-aware follow-up
Answers determine the next branch, so the interview adapts instead of following a static checklist.
Recommendation pressure
When useful, the assistant should say what it recommends and why, not merely list options.
Codebase awareness
If a codebase is available, the skill can explore files to ground questions in existing reality.
Interview tree
What are you building?
Define the actual goal.
What changes first?
Pick the entry point.
What can break?
Surface risks early.
What should I choose?
Recommend a path.
How the skill works
The value is in forcing decisions before implementation starts.
A good grilling session turns uncertainty into visible forks. It asks about ownership, scope, success criteria, fallback behavior, data shape, user experience, integration boundaries, testing, deployment, and the cost of being wrong.
- Ask exactly one meaningful question at a time.
- Wait for the user's answer before taking the next branch.
- Recommend a concrete answer when the tradeoff is clear enough.
Question categories
Grill Me Skill works best when the questions get specific fast.
What is in and out?
Clarifies the boundary so the plan does not expand while work is underway.
What is most likely to fail?
Names the fragile assumptions, unknown dependencies and hard-to-reverse choices.
Which option should we pick?
Moves from open-ended discussion to a defended recommendation with tradeoffs.
Use cases
The skill is useful before building, refactoring, shipping or delegating.
Feature planning
Pressure-test user flows, edge cases, data requirements, product scope and success measures.
Technical design
Question module boundaries, migration strategy, API contracts, performance and failure behavior.
Codebase changes
Use repository context to ask what existing patterns, tests and ownership boundaries affect the plan.
Decision review
Compare options and force a recommendation when the cost of indecision is higher than the risk.
Quick answers
Grill Me Skill FAQ
What is Grill Me Skill?
Grill Me Skill is an AI skill for interrogating a user about a plan, design, feature or implementation approach through focused, branching questions.
Why does it ask one question at a time?
One question at a time keeps the user from hand-waving through hard choices. Each answer becomes the basis for the next branch.
Does the skill only ask questions?
No. The referenced instructions also ask the assistant to recommend answers when useful, especially when a decision point has an obvious practical tradeoff.
Can it work with a codebase?
Yes. If codebase context is available, the skill can inspect relevant files so the questions are grounded in the actual project instead of abstract advice.