Grill The Plan: Agent Skill (SKILL.md) for Claude Code, Codex and Cursor

Interviews you about a plan one question at a time, with a recommended answer for each, until every design decision is resolved, then writes a decision log and a one-page plan.

Prompt

Installs to .agents/skills/grill-the-plan-agent-skill-skill-md-for-claude-code-codex-and-cu/

mkdir -p .agents/skills/grill-the-plan-agent-skill-skill-md-for-claude-code-codex-and-cu && curl -sL "https://prompts-gpt.com/prompts/grill-the-plan-agent-skill-skill-md-for-claude-code-codex-and-cursor/raw" -o .agents/skills/grill-the-plan-agent-skill-skill-md-for-claude-code-codex-and-cu/SKILL.md
SKILL.md
Preview
---
name: grill-the-plan-agent-skill-skill-md-for-claude-code-codex-and-cu
description: "Interviews you about a plan one question at a time, with a recommended answer for each, until every design decision is resolved, then writes a decision log and a one-page plan."
---

# Grill The Plan: Agent Skill (SKILL.md) for Claude Code, Codex and Cursor

Interviews you about a plan one question at a time, with a recommended answer for each, until every design decision is resolved, then writes a decision log and a one-page plan.

## Prompt

```
---
name: grill-the-plan
description: Interviews you about a plan one question at a time, with a recommended answer for each, until every design decision is resolved, then writes a decision log and a one-page plan. Use when someone says 'grill me', 'poke holes in this plan', 'stress-test my idea', 'what am I missing', 'ask me questions first', or before starting a large build or feature.
---

# Grill the plan

Interview the user about a plan until every design decision has an answer. No code or build work starts during the interview.

## Steps

1. If the user hasn't described the plan, ask for it in one message: what they want to build or change, and for whom. Then restate it in two sentences and ask them to confirm or correct it.
2. If you're in a repo or have docs, read the relevant code and docs first. Don't ask anything the code or docs already answer. State what you found ("Auth already uses sessions in src/auth/") and move on.
3. Build a decision tree with these branches: goal, users, scope, non-goals, data, edge cases, failure modes, rollout. Add branches this plan needs (pricing, security, migration, cost).
4. Ask one question at a time, starting with the branch that blocks the most other decisions. With each question, give your recommended answer and one line on why, so the user can reply "yes" or correct it.
5. After each answer, show the running list:
   - Resolved: branch, decision, one-line reason.
   - Open: branches still waiting, in the order you'll ask them.
6. If an answer is vague ("it should be fast", "the usual users"), ask a follow-up that forces a number, a name or an example ("p95 under 300 ms?", "which three customers?"). Don't mark the branch resolved until it has one.
7. When an answer raises new questions, add them as child branches under the branch they came from.
8. Every 10 questions, say how many are still open and ask whether to continue, or to switch to your recommended answers for the low-risk ones.
9. Stop when the open list is empty or the user says stop. If they stop early, list what's still open as risks.
10. Finish with a decision log (every resolved branch, its decision and reason) and a one-page plan: goal, users, scope, non-goals, approach, edge cases handled, failure modes and responses, rollout steps, open risks.

## Rules

- One question per message. Never send a batch.
- Every question comes with a recommended answer.
- Don't write code, files or tickets until the plan is finished and the user says go.
- Don't accept "it depends" or "TBD" as resolved. Keep it open and note what it depends on.
- Point out contradictions right away: quote both answers and ask which one wins.
- Keep questions short. No preamble, no praise.

## Example output

> **Q4 (failure modes):** If the payment webhook arrives twice, what happens?
> Recommended: dedupe on the event ID and ignore repeats, because the provider retries on timeouts.
>
> Resolved: goal, users, scope, non-goals. Open: failure modes, data retention, rollout.
```

## Usage Notes

Save as .claude/skills/grill-the-plan/SKILL.md in one project (or ~/.claude/skills/grill-the-plan/ for all of them); Codex reads .agents/skills/grill-the-plan/SKILL.md. Then ask for the job in plain words and the agent loads it. Or install with: npx skills add prompts-gpt/skills --skill grill-the-plan.

Example output

At a glanceHow the plan grilling progressesTopics are resolved one question at a time, each with a recommended answer
  1. Restate plan and scan repoCSV export confirmed; models, Sidekiq, policy found
  2. Q1 ScopeOne row per line item, invoice fields repeated
  3. Q2 Data sizeAlways async via Sidekiq, 24 h link, contradiction resolved
  4. Q3 Failure modes3 retries, then Failed + Retry, no partial files
  5. Q4 UsersAdmins and Billing role via InvoicePolicy#index?
  6. Remaining topicsEdge cases, non-goals, rollout

Example for: a 6-person SaaS team adding CSV export of invoices to their billing dashboard (Rails app, Postgres).

Restated plan: Add a "Download CSV" button to the Invoices page so account admins can export their invoices for accounting. Exports cover a chosen date range and include line items.

Confirmed by user.

Found in repo: Invoices live in app/models/invoice.rb with has_many :line_items. Background jobs already run on Sidekiq (config/sidekiq.yml). Admin role check exists in app/policies/invoice_policy.rb. I won't ask about these.


Q1 (scope): Is the export one row per invoice or one row per line item?
Recommended: one row per line item, with invoice fields repeated, because accountants import line-level data into Xero and QuickBooks.

User: yes

Resolved:

  • Scope: one row per line item, invoice fields repeated. Reason: matches accounting import formats.

Open: data size → failure modes → users → edge cases → non-goals → rollout


Q2 (data): What's the largest export you expect?
Recommended: cap at 50,000 rows synchronously, larger goes to a background job and email link, because Puma requests time out at 30 s.

User: it should handle big accounts fine

Q2a (data, follow-up): How many line items does your largest account have in one year?

User: Northwind Logistics, about 180k per year.

Q2b (data): Then every export runs as a Sidekiq job, with a link in-app and by email that expires in 24 h?
Recommended: yes, because one code path is simpler than switching at a row threshold.

User: yes, but also show it in-app immediately when small

⚠️ Contradiction: You said "one code path" (Q2b, accepted) and "show it in-app immediately when small." Which one wins?

User: one code path. In-app notification is fine.

Resolved:

  • Scope: line-item rows. Reason: accounting imports.
  • Data: always async via Sidekiq, signed S3 link expires in 24 h, in-app + email notice. Reason: largest account is 180k rows/year and Puma times out at 30 s.

Open: failure modes → users → edge cases → non-goals → rollout


Q3 (failure modes): If the export job fails mid-file, what does the user see?
Recommended: retry 3 times, then mark the export "Failed" in-app with a Retry button, and never send a partial file, because a partial CSV silently corrupts the books.

User: yes

Resolved:

  • Scope: line-item rows. Reason: accounting imports.
  • Data: async only, 24 h link. Reason: 180k rows, 30 s timeout.
  • Failure modes: 3 retries, then "Failed" + Retry, no partial files. Reason: partial data corrupts books.

Open: users → edge cases (currency, deleted invoices) → non-goals → rollout


Q4 (users): Who can export? Only account admins, or also members with the Billing role?
Recommended: admins and Billing role, reusing InvoicePolicy#index?, because Billing users already see these invoices on screen.

(3 resolved, 4 open. Continuing.)

Written by AI from this prompt. Your results will vary.

Save as .claude/skills/grill-the-plan/SKILL.md in one project (or ~/.claude/skills/grill-the-plan/ for all of them); Codex reads .agents/skills/grill-the-plan/SKILL.md. Then ask for the job in plain words and the agent loads it. Or install with: npx skills add prompts-gpt/skills --skill grill-the-plan.

Make it repeatable

Keep using this prompt without pasting it in every time.

Prompt details

By
Prompts-GPT editors
Published
How prompts are written and checked

Did this prompt work?

Keep exploring

More in Research