Matt Pocock's /grill-me skill: the AI questions you about the plan before it writes code
Most bad agent output starts with a misread request. /grill-me makes the agent interview you round by round over a decision tree, with a recommended answer for each question, before any work starts.
Visit site Claude Code / Codex + mattpocock/skills ↗- Pricing
- Free
- Region
- Global
- Where it's from
- Overseas
- Added
- Oct 16, 2026
- Version we used
- mattpocock/skills (checked 2026-10-16)
- Categories
- CodingProductivity
- Original author
- Matt Pocock @mattpocockuk ↗
- Our test
- Not tested by us
- Cost / time
- The skills are free and open source; you pay only your agent's usage. Expect a few extra minutes of Q&A up front.
- Source
- GitHub ↗X @alex_verem ↗
The task
Before Claude Code, Codex or a similar agent implements a feature or plan, surface every unstated decision and silent assumption so it doesn't build the wrong thing.
Steps
- Install the pack. Claude Code: claude plugin install mattpocock-skills@claude-plugins-official. Codex: codex plugin marketplace add mattpocock/skills, then codex plugin add mattpocock-skills@mattpocock (Copilot and Gemini CLI steps are in the README).
- Describe your plan or idea to the agent in a few sentences, then type /grill-me.
- The agent asks in rounds: each round lists every question whose prerequisites are settled, numbered, each with its recommended answer; replying "yes" accepts it.
- Facts it can look up (files, tools, environment) it finds itself, optionally via a sub-agent, instead of asking you.
- It only acts once every branch of the decision tree is visited and you confirm shared understanding. Follow with /to-tickets to split the plan into tickets.
Prompt
/grill-me (Core instruction from the skill: "Interview the user relentlessly until you reach a shared understanding. Map this as a design tree… Word each question so 'yes' accepts your recommended answer.")
What the author says
Per @alex_verem, the pack aims to sort out what you want before any work begins, and /grill-me works for non-code tasks too. ~293k GitHub stars when we checked on 2026-10-16.
Our test
We checked the repo, install commands and the grill-me / grilling skill text, but did not run a full session on a real project — no first-hand result to report.
How to get in
Skills pack is free on GitHub. You need Claude Code, Codex, Copilot or Gemini CLI access (usually a paid/overseas account). Chinese-language agents can still install the official plugin.
Who it suits
Developers and PMs using coding agents on non-trivial features, refactors or product plans, and anyone who wants to be pressure-tested before writing a doc or making a call.
Pitfalls
- Skip it for tiny changes; grilling a one-line fix wastes time.
- Pick either the plugin or skills.sh install per agent; installing both gives you every skill twice (the README warns about this).
- The recommended answer isn't automatically right; don't just say yes to everything.
More in this category
- A demo video per docs page that flags itself when the UI changesVideoScreenplayClaude CodeTool free; uses your agent's quota◐ Partly testedOverseas
- AI backend code that guards against races, leaks and bad migrationsCodingbackendskillsClaude CodePostgreSQLFree; about 2× tokens per task◐ Partly testedOverseas
- An afternoon of AI work, a 3D action game you play in the browserCodingClaude CodeTripoJimengSkill free; Claude quota plus asset tools◐ Partly testedChina
- Keep a long-running AI-built project from turning into a messCodingClaudeCodexoil-delegateNeeds Claude and Codex plans◐ Partly testedChina
- Use a web app once and get an AI write-up of how it worksResearchREAClaude CodeChromeFree tools; needs Claude or Codex◐ Partly testedOverseas