grill-me

Interviews the user one question at a time to stress-test plans and design decisions.

9|Updated Jul 17, 2026
One-click install
npx skills add https://github.com/Yassimba/loom --skill grill-me-yassimba
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: grill-me
Source: https://github.com/Yassimba/loom/tree/main/skills/grill-me
Command: npx skills add https://github.com/Yassimba/loom --skill grill-me-yassimba

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Plans and design decisions often hide unexamined assumptions that surface late as costly rework. This Skill stress-tests an idea by relentlessly interviewing the user, mapping every decision as a design tree and asking only the questions whose prerequisites are already settled. ## Core Features & Use Cases - Design tree mapping: Models the discussion as a tree of dependent decisions and works the frontier of answerable questions one at a time. - Structured questioning: Presents each question with a title, body, and a recommended answer, waiting for feedback before continuing. - Fact-finding via sub-agents: Looks up facts from the filesystem or tools itself instead of burdening the user, reserving questions for actual decisions. - Domain modeling support: When the session defines project language or enduring architecture, it maintains CONTEXT.md glossaries and ADRs per the bundled reference. - Use Case: Before writing a spec for a new billing system, run a grilling session to expose every unresolved decision about pricing models, edge cases, and failure handling until nothing is silently assumed. ## Quick Start Use the grill-me skill to stress-test my plan for migrating the monolith to microservices.

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 design decision with an AI agent?

Invoke the grill-me skill with your plan or idea. The agent maps your decisions as a tree and asks one question at a time, starting with decisions whose prerequisites are already settled, until every branch is examined.

How does the grilling interview decide which questions to ask?

It maintains a design tree where each decision branches into dependent decisions. It only asks questions on the frontier, meaning questions whose prerequisite decisions are already resolved, and recomputes the frontier after each answer.

Does the skill ask the user for facts about the codebase?

No. When a question needs a fact from the filesystem or tools, the skill dispatches a sub-agent to find it. Only genuine decisions are put to the user, one at a time.

When does the skill create CONTEXT.md or ADR documents?

Only when the session defines project-specific domain language or settles an enduring architectural decision. It then follows the bundled domain-modeling reference to maintain glossaries and ADRs; otherwise the session stays conversational.

When should I not use a grilling-style interview?

Avoid it for well-understood tasks with no open design questions, since the interview adds overhead. It is designed for unresolved plans and ideas where assumptions need to be surfaced before acting.