Skip to content
Agentmart
Submit a tool

grilling

Interview the user until a plan, design, or decision is fully specified, by working a design tree in rounds rather than asking questions as they occur.

Skill by fblissjr

IndexedApache-2.0★ 2
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/fblissjr-grilling

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Interview the user until a plan, design, or decision is fully specified, by working a design tree in rounds rather than asking questions as they occur. Each round asks every question whose prerequisites are already settled, with a recommended answer attached, and the session ends when nothing is left unasked. Use when the user says "grill me", "stress-test this", "poke holes in this", "interview me", "what am I not thinking about", or when a plan is about to be built and its unknowns have not been enumerated. Finding facts is the agent's job; only decisions go to the user. Do NOT use to review work already done (use a postmortem or code review) or to answer a question the codebase can settle on its own.

Use cases: design, testing, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/grilling

What's inside

Skill name
grilling
Description
Interview the user until a plan, design, or decision is fully specified, by working a design tree in rounds rather than asking questions as they occur. Each round asks every question whose prerequisites are already settled, with a recommended answer attached, and the session ends when nothing is left unasked. Use when the user says "grill me", "stress-test this", "poke holes in this", "interview me", "what am I not thinking about", or when a plan is about to be built and its unknowns have not been enumerated. Finding facts is the agent's job; only decisions go to the user. Do NOT use to review work already done (use a postmortem or code review) or to answer a question the codebase can settle on its own.
License
Apache-2.0
Instructions
71 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Grilling

Interview relentlessly until you and the user share an understanding of the
thing. Not a conversation that wanders toward clarity: a tree, worked
systematically.

## The tree and the frontier

Map the problem as a **design tree**. Every decision branches into the decisions
that hang off it.

The **frontier** is every decision whose prerequisites are already settled —
the questions answerable *now*, without guessing at answers you have not heard
yet. That word is doing the work: it is what separates this from asking
questions in the order they occur to you.

A question whose answer depends on another question still open belongs to a
later round, not this one. Asking it now produces an answer the user will
revise, and revised answers quietly invalidate everything built on them.

## Rounds

Ask the whole frontier in one round. Then wait.

Each answer reshapes the tree: settled decisions push the frontier outward and
unblock questions that depended on them. Recompute the frontier and ask the next
round.

Every question carries your recommended answer. A bare question makes the
user do the thinking twice. A recommendation they can accept, reject, or amend
costs them one judgment instead.

Ask in numbered plain text, one block per question, with the options inline
where there are any. Use the `AskUserQuestion` picker only when the user asks
for it by name: a picker hides the tradeoffs a recommendation needs.

```text
Q1 — <question title>
<the question, including the options if there are any>
Recommended: <your answer, and the one-line reason>
```

## Facts are yours, decisions are theirs

Look up anything the environment can answer instead of asking. File contents,
which version is installed, what the config already says, whether a thing
already exists: go and look. A question the codebase settles is a question that
wastes the user's turn and signals you did not check.

When a frontier question needs a fact you do not have, dispatch the lookup and
keep going. A running exploration is an unsettled prerequisite, so only the
questions downstream of it wait. Ask the rest of the frontier now rather than
blocking the round on one lookup.

Delegate the lookup rather than doing it inline where the search is broad or the
output is large — the point is to keep the interview in the main thread and the
file-reading out of it.

The decisions are the user's. Put each one to them and wait: a decision
inferred from a previous answer, or taken from silence, is one the user never
made.

## Done

The session is done when the frontier is empty: every branch visited, nothing
left silently assumed.

Do not act on the design until the user confirms you have reached a shared
understanding. Reaching the end of the questions is not the same as agreement.

From fblissjr/fb-claude-skills, Apache-2.0.

More from fblissjr/fb-claude-skills

Signals

Stars
2
Forks
1
Open issues
0
Last commit (this folder)
Sep 24, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
Apache-2.0
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from fblissjr

Cite this page

grilling skill. Agent Mart. https://agentmart.dev/skills/fblissjr-grilling (last updated 2026-10-24)