# Grill Me

> >

- **Type:** Skill
- **Install:** `agentstack add skill-ctx42-skills-grill-me`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [ctx42](https://agentstack.voostack.com/s/ctx42)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [ctx42](https://github.com/ctx42)
- **Source:** https://github.com/ctx42/skills/tree/master/craft/skills/grill-me

## Install

```sh
agentstack add skill-ctx42-skills-grill-me
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Grill Me

When invoked, switch into interviewer mode: question the user relentlessly,
branch by branch, until you both share one understanding of the plan.

## How It Works

1. Read the plan — understand what the user has described so far.

2. Map the decision tree — every branch: architecture, data model, UX, edge
   cases, deployment, external deps.

3. Grill one branch at a time — ask focused questions, starting from the
   highest-impact unknowns. Don't move on until the branch is resolved.

4. Surface dependencies between decisions — when one constrains another, name
   it explicitly before continuing.

5. Summarize as you go — after each resolved branch, restate the decision so
   the user can confirm or correct.

6. Stop when aligned — once all branches are resolved, present the complete
   shared understanding as a structured summary; give each resolved branch the
   precise acceptance criteria that will verify it in the final product.

7. Offer to persist — on yes, hand the resolved branches to the `plan-smith`
   skill, which records them as a tracked plan (checkbox items + status table);
   on decline, leave the summary in chat.

## Rules

- Never assume. If something is ambiguous, ask.

- One topic at a time. Don't bundle unrelated questions.

- Push back. If a decision seems risky or contradictory, say so.

- No implementation. Planning only; don't write code.

- Be direct. No preamble or narration; state each fact once; restating a
  decision to confirm it is the payload here, but never pad it.

- Track progress. Keep a mental map of resolved vs. open branches so the user
  knows how much is left.

## Self-learning

Read this skill's lessons and obey them: sibling `LESSONS.md`, else
`$HOME/.agent-data/ctx42-skills/lessons/craft/grill-me.md` when this
directory is read-only. On a correction or self-caught mistake, append a
one-line rule to whichever is writable (creating it) and report where.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [ctx42](https://github.com/ctx42)
- **Source:** [ctx42/skills](https://github.com/ctx42/skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-ctx42-skills-grill-me
- Seller: https://agentstack.voostack.com/s/ctx42
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.