Coding prompts · How to use a skill
review-plan-grill-me-prompt
A "grill me" prompt that makes the AI interview you round by round about a plan, design or idea, with a recommended answer for every question, until nothing is left assumed. Use when the user says grill me, wants their plan stress-tested, asks to be challenged before coding, or wants requirements nailed down first.
Version 1.0 · Updated 2026-10-31
Install in Claude Code, Cursor or Codex (optional)
Only for AI coding tools that read skill folders. Run one line in a terminal, in your project folder. It downloads this skill's text files from TheSkillVault and nothing else; you can read them first with “View raw SKILL.md”.
Claude Code (all your projects)
curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/review-plan-grill-me-prompt/raw' -o ~/'.claude/skills/review-plan-grill-me-prompt/SKILL.md' && curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/review-plan-grill-me-prompt/files/LICENSE.txt' -o ~/'.claude/skills/review-plan-grill-me-prompt/LICENSE.txt'Cursor (this project)
curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/review-plan-grill-me-prompt/raw' -o '.cursor/skills/review-plan-grill-me-prompt/SKILL.md' && curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/review-plan-grill-me-prompt/files/LICENSE.txt' -o '.cursor/skills/review-plan-grill-me-prompt/LICENSE.txt'Codex and other Agent Skills tools (this project)
curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/review-plan-grill-me-prompt/raw' -o '.agents/skills/review-plan-grill-me-prompt/SKILL.md' && curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/review-plan-grill-me-prompt/files/LICENSE.txt' -o '.agents/skills/review-plan-grill-me-prompt/LICENSE.txt'
Or connect the whole library over MCP
Add TheSkillVault to Cursor or add this to your app's MCP settings (read-only; help):
{
"mcpServers": {
"theskillvault": {
"url": "https://theskillvault.ai/mcp"
}
}
}Review my plan: grill me (prompt)
The AI asks you hard questions about your plan before anyone writes code. It works through the plan as a tree of decisions and asks a few at a time. Each question comes with its recommended answer, so you can often just say "yes".
When to use
- You have a plan, feature idea or design and want it stress-tested.
- You want the AI to stop guessing and ask you first.
- Before handing work to an agent, so nothing important is left vague.
Steps
- Paste the prompt below, then describe your plan (or point the AI at your plan file).
- Answer each round. Reply "yes" to accept a recommendation, or give your own answer.
- Keep going until the AI says every branch is settled. Then ask it to sum up the agreed plan.
Prompt
Grill me relentlessly about the plan below until we reach a shared understanding.
1. Map the plan as a design tree: each decision branches into the decisions that depend on it.
2. Work in rounds. The "frontier" is every decision whose prerequisites are already settled.
Ask the whole frontier in one round, then wait for my answers.
3. Number each question and give your recommended answer. Format each one as:
Q1 - <short title>: <the question, with options if useful>
Recommended: <your answer>
Word each question so that "yes" accepts your recommendation.
4. After each round, recompute the frontier. A question that depends on another open question
belongs in a later round, not this one.
5. Finding facts is your job, not mine. If a question needs a fact from the code, files or tools,
look it up yourself (use a sub-agent if your tool has them) instead of asking me.
Keep asking the other frontier questions while you look.
6. Decisions are mine. Put each one to me and wait.
7. Stop when the frontier is empty: every branch visited, nothing silently assumed.
Then give me a short summary of every decision. Do not start building until I confirm.
PLAN:
{{DESCRIBE YOUR PLAN OR NAME THE FILE}}
How to adapt
- Using docs or a glossary? Add "Check each answer against docs/ and GLOSSARY.md and flag conflicts."
- Short on time? Add "Ask at most 5 questions per round."
- Works with any AI chat or coding agent. If your tool has no sub-agents, the AI looks things up itself instead.
Gotchas
- The AI may drift into building early. Remind it: "Still grilling, no code yet."
- A question you cannot answer yet is a sign to research first, not to guess.
- Do not paste passwords, API keys or private customer data into the plan. Describe them instead ("we store a Stripe key in .env").
Done when
- The AI reports an empty frontier and lists every decision.
- You have confirmed the summary before any code is written.
Example prompt
"Grill me on my plan to add team accounts to my app before you write any code."
---
Adapted from "grill-me" / "grilling" in mattpocock/skills by Matt Pocock (MIT). See LICENSE.txt.
Adapted from Matt Pocock (mattpocock/skills, 'grill-me' / 'grilling') (Copyright (c) 2026 Matt Pocock), MIT License. Adapted by TheSkillVault: wrapped as a paste-in prompt, plain-language steps, sub-agent step made optional for tools without sub-agents, added safety, how-to-adapt and done-when sections.
Licence: MIT License. Attribution when you reuse it: Based on work by Matt Pocock (mattpocock/skills, 'grill-me' / 'grilling') (Copyright (c) 2026 Matt Pocock, MIT License), https://github.com/mattpocock/skills/blob/a496f5513d4a4cf889bc48e189e579f4b0bdf6aa/skills/productivity/grilling/SKILL.md; adapted by TheSkillVault, https://theskillvault.ai