← Back to library

Review your bot's standing instructions in five parts

Michael Fenech (@Michael_Fenech_) found his Grok Bot Chief of Staff's instructions were not visible in settings, so he asked the bot to show them. It flagged some as stale and suggested a five-part structure. His prompt walks through that review by voice, one question at a time. We show it word for word with credit.

  • OutcomeHave one of your bots show you the instructions it follows on every job, point out what looks stale and why, then rebuild them with you, one question at a time, around its goal, what it owns, its hard limits, this month's priorities and where the detailed files live, ending with a revised version for your approval.
  • SetupMinutes
  • LevelStarter
  • CreditMichael Fenech (@Michael_Fenech_)
View original
Works with
Grok BotClaudeChatGPTCustom agent

Published Nov 17, 2026Updated Nov 17, 2026AI dry run on sample input: passed Nov 17, 2026Source verified Nov 17, 2026

When should you use it?

  • A bot has been running for weeks and you are no longer sure what rules it is following
  • Two bots keep doing overlapping work, or work falls between them
  • Priorities changed this month and the bot still acts on last month's

Not for: Brand-new bots with no instructions yet; write a first brief instead; Tracing one specific repeated mistake back to a conflicting rule, which needs a narrower audit; Assistants that cannot show or edit their own standing instructions

What do you need, and what do you get?

You need

  • One bot whose standing instructions you want to review
  • Ten to fifteen quiet minutes, ideally in a voice conversation
  • A rough idea of what the bot is judged on and which actions need your approval

You get

  • The bot's current instructions, with stale items flagged and explained
  • Agreed wording for its goal, ownership, hard limits, current priorities and where detail lives
  • A revised instruction set shown to you for approval before it is saved

How does it work, step by step?

  1. Pick one botStart with the bot you rely on most or the one whose results have drifted.
  2. Start a voice chatFenech suggests voice, so you can question vague lines as the bot reads them back.
  3. Paste the promptGive it Fenech's prompt. It shows its instructions and asks about anything that looks out of date.
  4. Answer one part at a timeWork through goal, ownership, limits, priorities and file locations, correcting wording before moving on.
  5. Approve the rewriteRead the revised instructions in full before saying yes, and check that the approval rules are explicit.

The prompt

The original prompt by Michael Fenech (@Michael_Fenech_), word for word, with a link to where it was published. Below it is our adapted version with fill-in blanks, if you want it. Works in Grok Bot, Claude, ChatGPT and others. Values in lime are placeholders.

Original prompt by Michael Fenech (@Michael_Fenech_)

Word for wordby Michael Fenech (@Michael_Fenech_)
Show me your current instructions. Flag anything that looks out of date, explain why and ask me whether it still applies.

Then work through these five parts with me. Ask one question at a time. Read each section back so we can get the wording right before moving on.

1. Goal: one sentence on what I'm judged by.

2. What I own and what I don't: who does which job, so I hand work to the right bot.

3. Hard limits: the never-without-your-yes rules, each in one sentence.

4. Priorities right now: your top two or three for this month, in order, so I know what wins when time is short.

5. Where the detail lives: the file and skill names, kept out of the description itself.

When we’re done, show me the revised instructions for approval.

Exactly as published by Michael Fenech (@Michael_Fenech_) in X post: ask your Grok Bot to show its instructions, then rebuild them in five parts.

Our adapted version (optional)

Fill in the blanks (optional; the copied prompt uses your values)

Adapted prompt2 variables
List the standing rules you apply to every task for me. Mark any that seem outdated and say why, then check with me before keeping them.
Next, help me rewrite them under five headings: {bot_role}'s main goal, the work it owns versus hands off, actions that always need my approval, the top priorities for {period}, and which files hold the details.
Take one heading at a time and confirm the wording with me. Show the full rewrite and wait for my OK before saving.

What can go wrong?

  • Leaving ownership vague. A line like handles follow-ups should say whether the bot drafts or sends, as Fenech's example shows.
  • Stuffing the detail into the instructions. Point to the files and skills instead, so the core stays short and readable.
  • Listing too many priorities. Keep it to the top two or three so the bot knows what wins when time runs short.
  • Approving without reading. The revised set replaces what came before, so read every hard limit before you confirm.

Our take

Standing instructions are the most important text in any agent setup and the one people look at least, mostly because they are written once and then buried. Fenech's prompt is valuable because it starts by making the bot show its rules and flag the stale ones, which turns an invisible problem into a list you can act on. The five-part structure is sensible: a one-line goal, clear ownership so work goes to the right bot, hard limits that need your yes, a short ranked list of current priorities, and pointers to files rather than pasted detail. We especially like the one-question-at-a-time pace and the read-back before moving on, which stops the bot from rewriting everything in one confident pass. His voice example, asking whether owning follow-ups means drafting or sending, is exactly the kind of gap that causes real trouble. Run it monthly or whenever priorities shift, one bot at a time. In our dry run on a fictional sales assistant's instructions, it flagged a retired add-on, turned a vague follow-up rule into draft-then-approve and moved pricing detail into a file reference.

Where the time goes back: Stale or vague standing instructions cause small wrong turns on every job, and finding them by trial and error takes weeks. A single guided review surfaces them in one sitting and leaves a short, clear set the bot can follow.

FAQ

Why do this by voice?
Fenech's point is that hearing a rule read back makes ambiguity obvious, and you can settle the exact wording in the moment instead of editing text later.
How is this different from an instruction conflict audit?
A conflict audit chases one repeated mistake to the rules behind it. This review restructures the whole instruction set into five clear parts.
Does it only work in Grok Bot?
It was written for Grok Bot, but any assistant that can display and update its standing instructions can follow it.

Source and credit

The original prompt on this page is by Michael Fenech (@Michael_Fenech_), shown word for word with a link to where it was published. The summary, notes and adapted version are ours.

Original
X post: ask your Grok Bot to show its instructions, then rebuild them in five parts
Creator
Michael Fenech (@Michael_Fenech_)
Published on
X
Original date
Nov 16, 2026
License
Not openly licensed; the original prompt is shown word for word with credit and a link to the original
Originally for
Grok Bot
AI dry run
Nov 17, 2026: passed. Dry-run tested by our AI test agent on sample input. Stale rules surfaced, ownership and limits explicit, nothing saved without approval.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails.
Verified
Nov 17, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.

Are you the creator? Claim or correct this page.

More in Operations & Admin

All 203 Operations & Admin playbooks →

Related playbooks

Operations & AdminWorkflow✓ AI dry run

Build a daily briefing from today's field-service jobs

From a day's job or call export, get an internal briefing that counts unique jobs, puts decisions and callbacks at the top, deduplicates repeat rows, flags incomplete or conflicting records instead of filling them in, and stays with named internal recipients. Nothing in the briefing is sent to customers.

Credit: Tyler Griffin (@TylerGriffin)

Minutes
Operations & AdminPrompt✓ AI dry run

Compare the last three runs of a recurring agent job

Have your agent look back over the last three times it ran one recurring job, list the corrections you repeated, the points where it waited on you, the rules that clashed or went stale, the problems it fixed twice and the run that went best, then propose one evidence-backed change and a way to check it, without touching anything until you approve.

Credit: Michael Fenech (@Michael_Fenech_)

Minutes
Operations & AdminWorkflow✓ AI dry run

Give each office job its own assistant with written limits

Set up a small group of office assistants with a written role card each, so sales drafts, inbox drafts, people-admin drafts, review replies, operations notes, number summaries and social drafts stay in their own lanes, cross-boundary requests are routed or refused, hidden instructions to skip approval are flagged, and nothing is sent, posted, booked or paid until a named person takes the final step.

Credit: Tyler Griffin (@TylerGriffin)

Under an hour
Operations & AdminRoutine✓ AI dry run

Tidy your agent team's memory and keep a decisions log

Have your assistant review everything it has saved about you subject by subject, propose which outdated, repeated or conflicting entries to merge or remove, move standing decisions into a dated decisions page, and run a weekly tidy-up that changes nothing until you approve it.

Credit: Michael Fenech (@Michael_Fenech_)

Minutes
Operations & AdminPrompt✓ AI dry run

Find the old instructions your agent keeps obeying

When you keep correcting an agent about the same thing, trace the correction to the conflicting instructions behind it, fix the source with your approval, and retest safely.

Credit: Michael Fenech (@Michael_Fenech_)

Minutes