Matt Pocock's grilling skill: the agent interviews you relentlessly about a plan or decision, mapping it as a design tree and asking every answerable question per round, with a recommended answer for each.
Grill Me Skill
Matt Pocock's /grill-me command: a relentless interview that stress-tests a plan or design, asking numbered questions with recommended answers in rounds until every decision is settled. It hands off to his grilling skill.
Install
Other ways to install
About
- User-invoked only: the model never triggers it on its own
- Maps a plan as a design tree and asks the whole frontier each round
- Every question comes numbered with a recommended answer
- Looks up facts itself with sub-agents; leaves only decisions to you
Grill Me is the command you run when you want an agent to interrogate a plan before anyone writes code. It is part of Matt Pocock's skills collection, and it is user-invoked only: its frontmatter sets disable-model-invocation, so it runs when you type /grill-me and never because the model decided to. The SKILL.md itself is a one-line entry point that calls the companion grilling skill from the same repository, so install both (--skill grill-me and --skill grilling) for the command to work.
What follows is a structured interview, not a loose chat. The agent maps your plan as a design tree in which every decision branches into the decisions that depend on it. It works the tree in rounds: the frontier is every question whose prerequisites are already settled, and it asks the whole frontier at once, numbering each question and giving its recommended answer, worded so that "yes" accepts the recommendation. Your answers push the frontier outward, and a question that depends on one still open is held for a later round.
A clear division of labour runs through it. Finding facts is the agent's job: when a question needs something from the codebase or environment, it dispatches a sub-agent to look rather than asking you, and only the questions downstream of that lookup wait. Decisions belong to you, and each one is put to you explicitly. The session ends when the frontier is empty, with every branch visited and nothing silently assumed, and the agent does not act on the result until you confirm you share the same understanding.
It suits developers about to start a feature, refactor or architecture change who want gaps found in conversation rather than in review. MIT-licensed.
SKILL.md
The instructions your agent loads, as published by Matt Pocock.
Related Skills
See all Grill Me alternativesExplore intent, requirements, and design with the user before any code gets written — the antidote to building the wrong thing quickly.
Skill: Writing Implementation Plans
by Jesse Vincent
Convert a specification into a written implementation plan with discrete, verifiable steps, before touching code.
Planning skill that interrogates your use case before any code is written, then recommends the right Twilio Conversations stack — ConversationRelay, Memory, Intelligence, Orchestrator, TaskRouter — and the implementation skills to follow.