agentleFS
fortunto2fortunto2/solo-factory/skills/grill/SKILL.md

solo-grill

Skill18 starsChanged 4 months agoMITView on GitHub

Relentless one-question-at-a-time interview that stress-tests a plan, design, or decision before anything gets built. Use when user says "grill me", "прожарь", "разбери мою идею", "stress-test this", "punch holes in this", or a plan still has soft spots. Also reachable by other skills that need an interview. Do NOT use for scoring an idea (/validate) or writing the implementation plan (/plan).

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    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. Distinctive:

    Look up facts; ask for decisions: If something can be settled by exploring the environment — filesystem, git history, the codebase, the web …

    line 20
  2. Distinctive:

    Do not act on the plan until the user confirms the shared understanding is reached.

    line 22
  3. Very specific:

    Before /plan — when the feature is still fuzzy.

    line 32
  4. Very specific:

    Before /validate — when the idea itself hasn't been pinned down enough to score.

    line 33
  5. Very specific:

    After /diagnose Phase 3 — to pressure-test which hypothesis to chase.

    line 35
  6. Distinctive:

    Asking what you could look up burns trust: Check the repo, the config, the git log first.

    line 40
  7. Distinctive:

    Start building the moment the user answers the last question.

    line 45

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “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 — the interview primitive

Interview the user relentlessly about every aspect of the thing until you reach a shared understanding. Walk down each branch of the decision tree, resolving dependencies between decisions one by one. For each question, give your own recommended answer.

Ask one question at a time, waiting for the answer before the next. A batch of questions is bewildering and destroys the dependency order that makes the interview converge.

Look up facts; ask for decisions. If something can be settled by exploring the environment — filesystem, git history, the codebase, the web — go and settle it. The decisions are the user's: put each one to them and wait.

Do not act on the plan until the user confirms the shared understanding is reached.

Why one at a time

The mental model is a decision tree. An early answer reshapes which questions come next — a firehose of parallel questions throws away that structure and produces answers to questions that turned out not to matter.

Composing

This is a primitive, not a workflow. Reach for it from inside other work whenever a plan needs sharpening:

  • Before /plan — when the feature is still fuzzy. /plan deliberately asks zero questions and researches code instead, so grill first when the decisions, not the code, are what's missing.
  • Before /validate — when the idea itself hasn't been pinned down enough to score.
  • Inside /stream — when a STREAM layer needs the user's judgement rather than analysis.
  • After /diagnose Phase 3 — to pressure-test which hypothesis to chase.

Gotchas

  • Recommendations are the point. A bare question makes the user do all the work; a question with your recommended answer lets them correct a draft, which is far faster.
  • Asking what you could look up burns trust. Check the repo, the config, the git log first.

Don't

  • Batch questions "to save time" — it costs more.
  • Start building the moment the user answers the last question; confirm the shared understanding first.

Adapted from grilling by Matt Pocock (MIT, Copyright (c) 2026 Matt Pocock). See THIRD-PARTY.md.

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.