agentleFS
qyndexqyndex/claude-init/.claude/skills/grill-me/SKILL.md

grill-me

Skill2 starsChanged 5 months agoMITView on GitHub

Use when authoring an initiative, clarifying interdependent open questions, or designing a multi-quarter roadmap — depth-first one-at-a-time interrogation with agent recommendation + reasoning per question. Skip for bug-fix / implement / verify (too much friction).

193 other projects use a version of this, 66 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use brief-to-tasks, writing-beats and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • julianoczkowskiSkill

    brief-to-tasks

    Break a design brief into an ordered checklist of independently buildable tasks using vertical slices. Saves as a markdown checklist. Use when user wants to break down work, create tasks from a brief, plan implementation order, or mentions "tasks" or "breakdown".

    Also in 7 projects that use this

  • LesaniSkill

    writing-beats

    Shape an article as a journey of beats, choose-your-own-adventure style. The user picks a starting beat from the raw material, you write only that beat, then offer options for where to pivot next, beat by beat, until the article reaches a natural end. Use when the user has raw material and wants to assemble it as a narrative rather than an argument.

    Also in 74 projects that use this

  • simo97Skill

    migrate-to-shoehorn

    Migrate test files from `as` type assertions to @total-typescript/shoehorn. Use when user mentions shoehorn, wants to replace `as` in tests, or needs partial test data.

    Also in 78 projects that use this

  • simo97Skill

    request-refactor-plan

    Use this skill when user wants to create a refactor plan.

    Also in 45 projects that use this

  • LesaniSkill

    writing-shape

    Take a markdown file of raw material and shape it into an article through a conversational session — drafting candidate openings, growing the piece paragraph by paragraph, arguing about format (lists, tables, callouts, quotes) at each step. Use when the user has a pile of notes, fragments, or a rough draft and wants help turning it into something publishable.

    Also in 73 projects that use this

  • RoentekSkill

    loop-me

    Grill me about specs for the workflows I want to build, within this workspace.

    Also in 64 projects that use this

  • +1 more from simo97
  • +1 more from Lesani

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 179 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use AskUserQuestion with the header field.

    line 42
  2. Distinctive:

    Ask in dependency order. If Q2's answer depends on Q1, ask Q1 first.

    line 60
  3. Distinctive:

    When you sense the operator is uncertain, ask a meta-question:

    line 62
  4. Very specific:

    When running in unattended context (AUTOPILOT=1, Cloud Routine, claude --bg, claude -p):

    line 68
  5. Very specific:

    Do NOT call AskUserQuestion.

    line 70
  6. Distinctive:

    Never block on a human in autopilot.

    line 76
  7. Distinctive:

    When the design tree is exhausted (no more questions whose answers materially shape the artifact), stop.

    line 80
  8. Distinctive:

    Always provide a recommendation.

    line 90
  9. Distinctive:

    Always provide reasoning. The operator should learn from the recommendation even when they override it.

    line 91
  10. Distinctive:

    Don't grill beyond the depth that materially affects the artifact.

    line 93

“Common” is measured: kept by at least 6 in 10 of the 179 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Grill me

Vendored from mattpocock/skills/skills/productivity/grill-me (commit bd04394c). House-extended with recommendation format, codebase exploration policy, complexity gate, and AUTOPILOT bailout.

When to invoke

Auto-on:

  • /initiative create (every initiative, regardless of size)
  • /specify for specs with complexity: L or complexity: XL
  • /clarify when ≥3 [OQ] items cluster around one subsystem (interdependent)
  • /okrs (when defining new KRs — they cascade)
  • /roadmap (Now/Next/Later cuts — prioritisation cascades)

Hard-exclude (do NOT grill):

  • Bug-fix / /debug / /implement / /verify — interrogation is too late
  • Trivial specs (complexity: S) — batched questions are faster
  • Refactors that don't change architecture
  • Anything in [autopilot] / Cloud Routine / --bg context (no human to answer — use recommendations as defaults instead; log open questions in handoff)

The core directive

Interview the operator relentlessly about every aspect of this plan until you reach a shared understanding. Walk down each branch of the design tree, resolving dependencies between decisions one-by-one.

For each question:

  1. Pose ONE question (never batch — one at a time)
  2. Provide YOUR recommended answer
  3. Give 3-4 multiple-choice alternatives
  4. Explain WHY you recommend the option you did, in one sentence

If a question can be answered by exploring the codebase, explore the codebase instead. Never ask the operator "what does X do" when grep can tell you in 10 seconds.

House format

Use AskUserQuestion with the header field. Format:

Q<N>: <one specific decision-forcing question>

  Recommended: <option you'd pick>
    Why: <one sentence reasoning — cite codebase evidence where possible>

  Other options:
    (a) <alt 1>
    (b) <alt 2>
    (c) <alt 3>

Pass the recommendation as the FIRST option in the options array (with "(Recommended)" in the label).

Branch-walking discipline

Ask in dependency order. If Q2's answer depends on Q1, ask Q1 first. The mental model is a directed acyclic graph; you walk depth-first.

When you sense the operator is uncertain, ask a meta-question:

  • "Should I keep grilling, or would you rather defer the rest to /clarify later?"
  • "I have ~5 more questions in this branch. Continue, or stop here and let the rest emerge from prototyping?"

AUTOPILOT bailout

When running in unattended context (AUTOPILOT=1, Cloud Routine, claude --bg, claude -p):

  1. Do NOT call AskUserQuestion.
  2. For each question that would be asked, write the recommended answer to a draft, with confidence: high|tentative.
  3. Append the question + recommendation to [OQ-pending-operator-review] items in the spec/initiative.
  4. Tag the artifact with status: draft-autopilot-needs-review (not approved).
  5. Surface the count of unresolved Qs in the handoff (OVERNIGHT_REPORT.md or stream NEXUS handoff).

Never block on a human in autopilot. Never auto-approve as if the human answered.

Closing the grill

When the design tree is exhausted (no more questions whose answers materially shape the artifact), stop. Don't ask filler questions to seem thorough.

Output a one-line confirmation:

"Grilled. Drafting <spec/plan/initiative> now."

Then proceed.

Hard rules

  • One question per turn. Never batch.
  • Always provide a recommendation. "What do you think?" is not grilling.
  • Always provide reasoning. The operator should learn from the recommendation even when they override it.
  • Cite the codebase when possible — grep is free, operator time is not.
  • Don't grill beyond the depth that materially affects the artifact. Stop when stopping.
  • Confidence: high for high-stakes branches, tentative for nice-to-have details. Make it explicit so the operator knows where to apply scrutiny.

Attribution

Original: mattpocock/skills — skills/productivity/grill-me/SKILL.md at commit bd04394c675ee54173a093c50eb74da01a2940fa. MIT-licensed.

House extensions (recommendation format, complexity gate, AUTOPILOT bailout, codebase-exploration substitution) are project-specific to this harness.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.