← Back to library

Check which of your agents should be merged or retired

Michael Fenech (@Michael_Fenech_) shared an audit for Grok Bot teams that asks a question most owners skip before adding another agent: should some of the current ones exist at all? It ends in a plan for retiring agents safely. We explain how to use it and show his prompt word for word with credit.

  • OutcomeHave your lead agent map every agent on your team with its jobs, routines, skills, connections and recent activity, find overlap and unused output, recommend keep, absorb, retire or hand back to you for each, and draft a safe wind-down plan covering routines, open work, saved instructions, files and sign-ins, with nothing removed until you approve.
  • SetupUnder an hour
  • LevelIntermediate
  • CreditMichael Fenech (@Michael_Fenech_)
View original
Works with
Grok BotClaude CoworkChatGPT agentCustom agent

Published Nov 3, 2026Updated Nov 3, 2026AI dry run on sample input: passed Nov 3, 2026Source verified Nov 3, 2026

When should you use it?

  • Your agent team grew one agent at a time and nobody has reviewed it since
  • Before you create another agent for a job an existing one might cover
  • Usage is climbing and you suspect some agents produce work nobody reads

Not for: A setup with one or two agents you check every day; Platforms where the lead agent cannot see the other agents' routines or history; Deciding whether one overloaded agent should be split; that is a different audit

What do you need, and what do you get?

You need

  • A lead agent that can read every agent's instructions, routines, skills and connections
  • Recent activity or task history for each agent
  • Your own view of which outputs you actually use

You get

  • A team map with each agent's jobs, schedules, skills, connections and recent work
  • Overlaps and outputs nobody uses, with examples
  • A keep, absorb, retire or human verdict per agent with reasons and expected benefit
  • A wind-down checklist for each agent you approve removing

How does it work, step by step?

  1. Run it read-onlyGive Fenech's prompt to your lead agent and confirm it will change nothing.
  2. Challenge the verdictsAsk for the evidence behind every absorb or retire call, such as last use or a duplicate job.
  3. Move the work firstFor approved changes, transfer routines and open tasks to their new owner and confirm they ran.
  4. Keep what is worth keepingSave instructions, memory notes and lessons into the agent that takes over the work.
  5. Clean up accessRemove leftover files, connections and sign-ins only after the new owner is working.

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 Cowork, ChatGPT agent and others. Values in lime are placeholders.

Original prompt by Michael Fenech (@Michael_Fenech_)

Word for wordby Michael Fenech (@Michael_Fenech_)
Run a Bot retirement audit across my entire Grok Bot team.

1: Map the team. List every Bot, its responsibilities, active routines, skills, integrations and recent activity.

2: Find the overlap. Identify Bots doing similar jobs, duplicating routines or producing outputs nobody uses.

3: Check actual value. For each Bot, assess what it accomplishes, how often it's needed and whether another Bot could handle the work.

4: Make a decision. Recommend KEEP (valuable and distinct), ABSORB (merge its responsibilities into another Bot), RETIRE (no longer needed) or HUMAN (the task is better handled by me).

5: Build a safe retirement plan. Before removing anything, identify routines to stop, responsibilities to transfer, memory or instructions worth preserving, and any shared files, integrations or sign-ins that need attention.

6: Show me the result. Create a table with each Bot, your recommendation, reasoning, expected benefit and the actions required.

Don't delete Bots, disable routines or change permissions without my approval.

Exactly as published by Michael Fenech (@Michael_Fenech_) in X post: a Bot retirement audit to run before adding another Grok Bot.

Our adapted version (optional)

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

Adapted prompt2 variables
Review every agent on my team on {platform} and tell me which ones I still need. Change nothing while you do it.

For each agent, list its jobs, schedules, skills, connected apps and what it produced in the last {period}. Point out agents whose work overlaps and outputs nobody has opened.
Give each agent one verdict: keep, fold into another agent (name it), retire, or return the task to me. Explain why and what I gain.
For anything you suggest removing, list the routines to stop, open work and saved instructions to move, and files, connections or logins to clean up afterward.
Before calling an agent idle, check for monthly, quarterly or yearly jobs. Wait for my approval before any change.

What can go wrong?

  • Hiding or pausing an agent can leave its schedules running. Stop routines explicitly and check they stopped.
  • Deleting an agent can leave files and logins behind on shared machines. Clean those up as a separate step.
  • Rarely used agents may still cover monthly or yearly jobs. Look at the full calendar before retiring one.
  • A human verdict means the task returns to you. Make sure you have time for it before agreeing.

Our take

Most advice about agent teams is about adding: another specialist, another routine, another connection. Fenech's audit runs the other direction and asks what you could remove, which is the more useful question once a team has grown by accident. The four verdicts are well chosen, especially the option to hand a task back to the owner, an honest admission that some jobs are faster done by a person than supervised. The wind-down step is what makes it safe. Retiring an agent is not only switching it off; its schedules, open tasks, saved instructions and shared logins all need somewhere to go. His note that hiding an agent leaves its routines running is a practical warning worth repeating. Our caution is to check the calendar before trusting an idle verdict, since quarterly or year-end jobs look unused for most of the year. In our dry run on a fictional five-agent bookkeeping practice, the audit kept two agents, folded a second inbox agent into the first, retired a weekly report nobody had opened in six weeks, returned vendor calls to the owner, and listed two routines and one shared login to deal with before removal.

Where the time goes back: You stop paying in usage and attention for agents that repeat each other or produce reports nobody opens, and you avoid adding new agents to cover gaps the existing team could absorb.

FAQ

How is this different from checking whether an agent should be split?
A split audit looks for overload in one agent. This one looks across the team for agents that overlap, sit idle or produce work you no longer need.
Will it delete anything?
It should not. The last line of the prompt forbids deleting agents, disabling routines or changing permissions without your approval.
Does it only work in Grok Bot?
It was written for Grok Bot, but any lead agent that can see your other agents' setup, for example in Claude Cowork or a custom framework, can run 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: a Bot retirement audit to run before adding another Grok Bot
Creator
Michael Fenech (@Michael_Fenech_)
Published on
X
Original date
Nov 2, 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 3, 2026: passed. Dry-run tested by our AI test agent on sample input. Found the duplicate and the unused report, used all four verdicts, listed routines and the shared login to handle, no changes made.
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 3, 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 & AdminPrompt✓ AI dry run

Audit how you manage your agent team, not only the agents

Have your chief-of-staff agent review the last 30 days of how you delegated, clarified, interrupted, corrected, assigned, approved and followed up across your agent team, score eight management areas with evidence, list your three biggest weaknesses, and draft a seven-day improvement plan, without changing any bot, routine or instructions until you approve.

Credit: Michael Fenech (@Michael_Fenech_)

Under an hour
Operations & AdminPrompt✓ AI dry run

Brief an agent with nine fields, not a how-to

Before you hand a job to an agent, fill in nine short fields that name the result, the sources it may use, the deadline, what it may do alone, what needs your say-so, what it must never do, how to prove the work is finished, what to do when it gets stuck, and where to put the deliverable, then add a line that tells it to pick the simplest reliable method.

Credit: Michael Fenech (@Michael_Fenech_)

Minutes
Operations & AdminPrompt✓ AI dry run

Brief your agent on the outcome, not every step

Hand a new job to your lead agent by describing the finished result instead of the steps, and have it confirm what success looks like, ask only for context it cannot find itself, pick the simplest dependable approach, state what it will do alone and what needs your approval, and check the result against your goal before calling it done.

Credit: Michael Fenech (@Michael_Fenech_)

Minutes
Operations & AdminPrompt✓ AI dry run

Give every agent routine a clear point where it stops

Have your lead agent review every recurring job, monitor and follow-up your agents run, check each against four stop rules (what counts as done, what should hand control back to you, how many retries or days are too many, and how a stop gets recorded), then propose fixes ranked by risk, changing nothing until you approve.

Credit: Michael Fenech (@Michael_Fenech_)

Under an hour
Operations & AdminPrompt✓ AI dry run

Have your agent trace the ripple effects of each decision

Add a standing rule to your project agent so that whenever you confirm a decision, it searches the plan, email, calendar and documents for the tasks, dates and promises that depended on the old choice, updates what it is allowed to change, drafts the rest for your approval and reports what moved and why.

Credit: Michael Fenech (@Michael_Fenech_)

Minutes