What does 'doing homework' mean for the Grill skill?

Before asking the user a question, the skill searches the codebase for existing patterns and researches technical trade-offs so it only asks about genuinely ambiguous points.

When should I use the Grill skill?

Use it whenever you have a plan, design, or architecture that needs a 'devil's advocate' to find gaps, contradictions, or hidden risks before implementation begins.

How does the decision tracker prevent context loss?

At the end of every response, the skill provides a compact status block showing which branches are resolved, pending, or blocked, ensuring the conversation stays on track.

How many questions will the skill ask at once?

To maintain focus and avoid overwhelming the user, Grill prioritizes the most critical dependencies and asks a maximum of two surgical questions per turn.

Can I use Grill for non-code planning?

Yes, it is effective for API contracts, database schemas, deployment strategies, and any project plan with a complex decision tree.

Grill - Design Stress Tester

Name: Grill - Design Stress Tester
Author: denniscalazans

bydenniscalazans

•

コラボレーションツール

Stress-tests technical plans and architectures through rigorous codebase analysis and surgical questioning to identify hidden risks.

Grill is a high-intensity interrogation skill for Claude Code that transforms the AI from a passive assistant into a prepared architectural interviewer. Instead of just accepting a plan at face value, Grill performs extensive reconnaissance on your existing codebase to find patterns and trade-offs before challenging your design. It maps out complex decision trees, surfaces contradictions, and forces a shared understanding by resolving dependencies one-by-one. It is specifically designed to minimize mental overhead for the developer by providing ranked decision tables and persistent status trackers, ensuring that second-order consequences are addressed and nothing falls through the cracks during the planning phase.

主な機能

01Ranked decision tables comparing options with pros, cons, and weighted recommendations

023 GitHub stars

03Automatic detection of contradictions, over-engineering, and hidden dependencies

04Branch-by-branch interrogation focused on resolving the critical decision path

05Autonomous codebase reconnaissance to answer basic questions before engaging the user

06Persistent status trackers to prevent context loss during long architecture reviews

ユースケース

01Conducting a deep-dive review of a migration plan to surface second-order consequences

02Challenging a proposed feature implementation to ensure it aligns with existing code patterns

03Vetting a new system architecture or database schema for potential scaling and maintenance issues

主な機能

01Ranked decision tables comparing options with pros, cons, and weighted recommendations

023 GitHub stars

03Automatic detection of contradictions, over-engineering, and hidden dependencies

04Branch-by-branch interrogation focused on resolving the critical decision path

05Autonomous codebase reconnaissance to answer basic questions before engaging the user

06Persistent status trackers to prevent context loss during long architecture reviews

ユースケース

01Conducting a deep-dive review of a migration plan to surface second-order consequences

02Challenging a proposed feature implementation to ensure it aligns with existing code patterns

03Vetting a new system architecture or database schema for potential scaling and maintenance issues