roast

Interrogate plans with one well-scoped question per turn.

Updated May 8, 2026
One-click install
npx skills add https://github.com/ace3/skills --skill roast-ace3
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: roast
Source: https://github.com/ace3/skills/tree/main/skills/roast
Command: npx skills add https://github.com/ace3/skills --skill roast-ace3

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Roast helps teams clarify complex plans by forcing a single, well-scoped question per turn, ensuring alignment before moving forward.

Core Features & Use Cases

  • Comprehensive plan interrogation: maintain an internal dependency map of resolved/open/blocked decisions.
  • Guided critique: push back with concrete recommendations and ask exactly one question per turn.
  • End-to-end planning support: suitable for design reviews, PRD drafting, architecture discussions, and plan stress-testing.

Quick Start

Provide the plan or task brief and let Roast respond with one precise question and a concrete recommendation.

Frequently Asked Questions about roast

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

FAQPage Schema
How do I stress-test an architecture plan before implementation?

You stress-test an architecture plan by subjecting it to a guided critique that pushes back with concrete recommendations and asks exactly one well-scoped question per turn. This enforces alignment and exposes structural flaws before development begins.

What is the best way to critique a PRD draft for missed requirements?

Critiquing a PRD draft for missed requirements requires a strict protocol that pushes back with a concrete recommendation and asks a single pointed question per turn. This ensures all complex plan dependencies are resolved before moving forward.

Can I use structured design reviews to resolve blocked decisions across codebases?

Yes, you can resolve blocked decisions across codebases by applying a strict interrogation protocol that maintains an internal dependency map of open and blocked items. It forces one pointed recommendation per question to ensure alignment.

How does a one-question-per-turn protocol improve decision-making for complex plans?

The one-question-per-turn protocol improves decision-making by preventing scattered discussions and forcing teams to address specific dependencies sequentially. It maintains an internal dependency map to ensure each open decision receives a single pointed recommendation.

Does guided plan interrogation work for stress-testing complex software engineering tasks?

Guided plan interrogation works for stress-testing complex software engineering tasks by enforcing a strict protocol that maintains an internal dependency map. It drills into the plan by requiring one well-scoped question and a single pointed recommendation per turn.