grill-me

Interviews the user in structured rounds to stress-test plans and decisions.

Updated Aug 19, 2026
One-click install
npx skills add https://github.com/nperepichka/Antigravity --skill grill-me-nperepichka
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: grill-me
Source: https://github.com/nperepichka/Antigravity/tree/main/config/skills/grill-me
Command: npx skills add https://github.com/nperepichka/Antigravity --skill grill-me-nperepichka

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Plans, decisions, and ideas often contain hidden assumptions and unresolved dependencies that only surface during implementation. This Skill systematically interrogates your thinking through structured questioning rounds until every decision branch is explicitly settled. ## Core Features & Use Cases - Design Tree Mapping: Models your plan as a tree of decisions where each choice branches into dependent sub-decisions. - Frontier-Based Rounds: Asks only the questions whose prerequisites are already settled, numbering each question and providing a recommended answer. - Autonomous Fact-Finding: Looks up environmental facts (filesystem, tools) itself instead of asking the user for information it can retrieve. - Use Case: Before starting a major refactor, invoke this Skill to be grilled on architecture choices, migration strategy, and rollback plans until no assumption remains unexamined. ## Quick Start Ask the agent to grill you on your migration plan using the grill-me skill and answer each numbered round of questions.

Frequently Asked Questions about grill-me

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I stress-test a plan before implementing it?▼

Invoke the grill-me skill and present your plan or decision. The agent maps it as a design tree and asks numbered questions in rounds, giving recommended answers, until every decision branch is explicitly settled.

What is a design tree in decision-making?▼

A design tree models a plan as decisions that branch into dependent sub-decisions. The skill works the tree in rounds, asking only the frontier questions whose prerequisites are already resolved, then recomputing the frontier after each answer.

Does the skill ask the user for facts about the environment?▼

No. The skill instructs the agent to look up facts from the filesystem or tools itself rather than asking the user. Only genuine decisions are put to the user; factual lookups never block unrelated questions.

When does a grill-me session end?▼

The session ends when the frontier is empty, meaning every branch of the design tree has been visited and nothing remains silently assumed. The agent does not act on the plan until the user confirms shared understanding.

When should I not use structured interrogation for decisions?▼

Avoid it for trivial, reversible choices where the overhead of multi-round questioning exceeds the risk. It is designed for consequential plans where hidden assumptions or unresolved dependencies could cause costly rework.