grilling

An agent skill by mattpocock, from mattpocock/skills. Tags: analysis, grilling, review, strategy, workflow.

What it does

Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add mattpocock/skills --skill grilling

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/mattpocock/skills
cp -r skills/skills/productivity/grilling ~/.claude/skills/grilling

Safety box score

Grade A, minimal risk: 5 of 100 risk points, rated 2026-10-16. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This is a conversational skill that only drives a structured Q&A process; it does not ship code, run commands, or modify the machine. The main consideration is behavioral: it may keep probing the user and may ask sub-agents to inspect local context for facts instead of asking the user.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	0 (none)	No scripts, shell commands, or code evaluation are present in `SKILL.md` or `agents/openai.yaml`.
Network access	0 (none)	The skill does not download from or send data to remote hosts.
Filesystem reach	0 (none)	It does not read, write, or delete files; it only describes how to ask questions and optionally consult a sub-agent.
Secrets & credentials	0 (none)	There is no instruction to access tokens, passwords, keychains, `.env` files, or other credentials.
Privilege & persistence	0 (none)	No sudo, installs, services, shell/profile edits, or other privilege-changing actions are requested.
Instruction hijack surface	1 (low)	The skill strongly directs agent behavior, including "dispatch a sub-agent" for environmental facts and "Don't block on it," but it also explicitly says "Do not act on it until the user confirms"; this is guidance, not malicious prompt injection.
Irreversible actions	0 (none)	It does not perform actions on the user's behalf; it only interviews the user and waits for confirmation.
Transparency	0 (none)	All shipped files are plain text and fully readable; there are no binaries, blobs, or downloaded installers.

Before installing

Confirm you want an agent that will keep asking follow-up questions until it believes every branch is settled.
Decide whether you're comfortable with the agent using sub-agents to inspect local context for facts instead of asking you.
No technical setup is needed; there are no scripts or system changes to review.

Rated 2026-10-16 by an automated review of 2 files from the skill's repository.

Source

Repository: mattpocock/skills (all skills from this repository)
Path: skills/productivity/grilling/SKILL.md
Branch: main
Updated: 2026-10-05

grill-me — A relentless interview to sharpen a plan or design.
maintain-verification-skill — Periodic pass that keeps a project's verification skill and feature map honest: parallel source readers per feature, one live session driving every feature.
principle-attack-the-premise — Apply when two or more fixes that share one premise have failed the same gate.
show-me-your-work — Keep a reviewable decision trail for long-running or unattended work: a TSV log with one row per decision (what, why, evidence, result).
why — Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds.
loop-me — Grill me about specs for the workflows I want to build, within this workspace.