Skip to content

Search setups and tools

Type a few words to search.

Browse all setups

Try a new tool behind an approval gate

Give a bot one new tool behind a gate: a sample run on safe data, the exact data it would touch, a rough cost, and an approval rule before real use.

For owners adding a new app or connector to a working bot who want proof it behaves before it touches real work · Grok Bot · Intermediate

Researched, not run yetWhat that means

Copy the prompt

Fill in the [brackets] after you paste it.

I want to give you access to [TOOL OR APP] for one job: [JOB]. Before any real use, do a sample run only.

1. Do the job once on this safe input: [SAFE SAMPLE DATA].
2. List the exact data, accounts and settings you touched, or would touch on a real run.
3. Give a rough cost: how much usage one real run like this takes, and how often you expect to run it.
4. List what could go wrong on a real run and how you would avoid each one.

Stop after the sample. Do not use the tool on real data, do not send, change or delete anything, and do not save an Always allow rule. Wait for me to review the sample and say what happens next.

Three ways to run this setup

  • Free

    Copy and paste it yourself

    Copy the prompt above into ChatGPT, Claude or Grok. It drafts. You approve.

    Go to the prompt
  • $9 once

    Complete Pack

    One payment, no membership. Opens the pack pages: Bot Safety Checklist, Second Brain Starter Vault, Voice File Starter, Tested Grok Bot prompt pack, Write it as code prompts.

    See the Complete Pack
  • Free to start, VIP $19 a month

    Save it as a bot in the guided setup app

    The app turns this setup into a bot with the draft-only rules. 3 setups build free; the rest are part of VIP. A human reviews a prompt or setup for you within 24 hours, up to 2 a month.

    Build it in the app

Want this one tested next? Vote for it.

What you get back

Illustrative. Your results will differ.

  • SAMPLE RUN: added one row to the test copy of the leads sheet. Would touch: Google Sheets app, Leads tab only.
  • ROUGH COST: about one routine run per new lead; expected 5 to 10 runs a week.
  • RULE ADDED: Ask first before creating or editing rows outside the Leads test tab.

3 steps

  1. Copy the prompt.
  2. Paste it into Grok Bot.
  3. Fill in the [brackets], then read the answer before you send or use it.

Quick answer

A new tool is new access, so earn trust in order: one sample run on safe data, a list of the exact data and accounts the tool would touch, a rough cost per run, and only then a narrow approval rule. In Grok Bot, Auto Review shows an approval card before connected-app and computer actions, and Custom Rules let you mark any action Ask first or Allow automatically.

How do you set it up?

Tools

  • Grok Bot with Auto Review on (it is on by default)
  • The new app or connector, installed but not yet used on real data
  • One safe sample input

Steps

  1. Pick one small real job for the new tool and one safe input: sample data, or a single record you can afford to fix if the run goes wrong.
  2. Ask for a sample run only. The bot shows what it would do, lists the exact data and accounts it would touch, and gives a rough usage cost per run, then stops.
  3. Check the sample: did it read only what you expected, and did the output match the job? If either answer is no, the tool is not ready.
  4. Set the approval rule before real use. In Grok Bot, add a narrow Ask first Custom Rule for the new tool's actions; xAI's docs say Ask first rules always stop matching actions for you.
  5. Run the real job once with the card on. Only after a clean sample and a clean first real run decide whether the action still needs your OK every time.
  6. Write the decision into the bot's instructions: the tool, the scope it may use, and what still needs approval. Chat corrections fade; written rules stay.

Where to paste this prompt

Grok Bot → paste the prompt into a new chat; save as a skill or routine only if you want this job to repeat

More ways to run it

Which chat tools work with this prompt?

This prompt needs routines and connectors, so it's written for:

  • Grok Bot

How sure is this?

ResearchedDrafts for you

Operations · Intermediate · Small business

Last updated: · Last reviewed:

Before you use this

What it is allowed to do: Drafts for you. Writes drafts (emails, posts, plans, replies). You review and send. Nothing goes out without you.

At a glance: time, cost and difficulty

At a glance

All outcome figures are estimates from cited sources, not independent verification by us.

What it is allowed to do
Drafts for you

Writes drafts (emails, posts, plans, replies). You review and send. Nothing goes out without you.

Reviewed Nov 3, 2026

Time saved (estimate)
Not measured yet
Difficulty
Intermediate
Tools needed
Grok Bot
Typical cost
Prices: Not stated in sources.

Basis (estimate): No timing claim: the sources describe approval and test controls, not time saved.

What usually goes wrong?

Where agents tend to stall on this job. More patterns in Where AI gets stuck.

  • Pressing Always allow on the first approval card. That saves a standing rule before the tool has earned it.
  • Writing a broad rule such as allow everything in the browser; xAI's docs warn that broad rules age badly because sites and tools change.
  • Skipping the sample because the tool looks simple, then discovering its real scope on a client record.

What tips make it smoother?

  • One new tool at a time. Two new tools in one week makes the sample runs impossible to read.
  • Keep the sample input boring. A sample that cannot hurt anyone is the whole point.
  • Write the allowed scope into the bot's instructions the day you approve it.

Where do the claims come from?

Summaries use public sources linked below, not independent verification. How setups are picked.

  • Read xAI’s write-up (docs.x.ai)

    xAI: Auto Review approval cards, Custom Rules (Ask first wins, up to 20 allow and 20 ask-first rules, 1,000 characters each), and the warning against broad allow rules (read 2026-11-03)

  • Read xAI’s write-up (docs.x.ai)

    xAI: Test before enabling: a test run performs real work, so use safe inputs and keep write actions behind approval (read 2026-11-03)

Common questions

What should a bot show me before I approve a new tool?
Three things: a sample run on safe data, the exact data and accounts it would touch, and a rough cost per run. If any of the three is missing, ask for it before approving.
When is Always allow reasonable?
After a clean sample run and a clean first real run, and only for a narrow action with a known scope. Saving it earlier means a standing rule you have not seen work.
Does an approval undo work already done?
No. xAI's docs are explicit: an approval controls the proposed action and does not reverse completed work, which is why the sample run comes first.

Free kit by email

Get the free kit by email

The starter pack and the Portable AI Office, sent to your inbox. New setups by email after that. Unsubscribe any time.

What to do next