grill-me

Runs a read-only design interview that challenges a proposed plan through a native subagent.

Updated Mar 26, 2026
One-click install
npx skills add https://github.com/sciman-top/skills-manager --skill grill-me-sciman-top
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: grill-me
Source: https://github.com/sciman-top/skills-manager/tree/main/overrides/patches/grill-me
Command: npx skills add https://github.com/sciman-top/skills-manager --skill grill-me-sciman-top

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Proposed plans and decisions often go unchallenged until implementation reveals flaws. This Skill launches an explicit, read-only design interview that interrogates a proposal one decision at a time, surfacing risks and assumptions before any code is written. ## Core Features & Use Cases - Native subagent dispatch: Delegates the interview to the host-native design-griller subagent via spawn_agent, keeping the parent conversation as the user-facing channel. - Read-only guarantee: The interview never edits repositories, writes ADRs, changes host configuration, or produces side effects. - Evidence-grounded routing: Requests combining interrogation with external evidence gathering are routed to the grill-with-docs closure instead of a plain interview. - Use Case: Before committing to a new architecture, ask the Skill to grill your migration plan; it returns one decision-changing question at a time, then a capsule of settled decisions, open risks, and assumptions. ## Quick Start Ask the assistant to use grill-me to challenge your proposed design or plan one decision at a time without modifying any files.

Frequently Asked Questions about grill-me

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

FAQPage Schema
How do I challenge a design proposal before implementing it?▼

Explicitly ask to grill or interrogate your plan, and the Skill spawns a design-griller subagent that asks one decision-changing question at a time. It returns settled decisions, open risks, and assumptions without modifying any files.

What is a read-only design interview?▼

A read-only design interview is a structured questioning session that probes a proposal's decisions without editing repositories, writing ADRs, changing configuration, or causing side effects. It only surfaces risks, assumptions, and settled decisions.

Can the design interview gather evidence from external docs?▼

Yes, requests combining interrogation with evidence gathering are routed through cold discovery to the grill-with-docs closure, which grounds each question in official docs or community practices. Evidence is gathered per question, never instead of asking it.

What happens if the native subagent cannot be spawned?▼

If spawn_agent is unavailable, fails, or returns no identifier, the Skill stops with a native_bridge_unavailable status naming the missing spawn evidence. It never simulates a child interview or falls back to a parent-side interview.

When should I not use this design interview skill?▼

Do not use it for ordinary implementation, debugging, or brainstorming sessions. It activates only when you explicitly ask to grill, challenge, or interrogate a proposed plan or decision.