When should you use it?
- You have a year of statements or billing emails and suspect forgotten recurring charges
- You want cancel candidates ranked by annual saving with evidence and fee notes
- You will approve every cancel; the agent must not cancel on its own
Not for: Letting the agent cancel, change payment methods or chat with support without your OK; A light monthly inventory from a short receipt list (use personal-subscription-audit-routine instead); Treating any dollar figure in the source post as a guaranteed saving
What do you need, and what do you get?
You need
- Last 12 months of bank or card statements you can upload
- Billing or renewal emails the agent can read, if available
- Which services you still use weekly, monthly or rarely
You get
- A table of recurring subscriptions showing price, renewal date and how to cancel
- Flags for duplicates, price increases and upcoming annual renewals
- Cancel candidates ranked by annual saving, each with evidence and fee notes
- Confirmation of which payments actually stopped after you approve cancels
How does it work, step by step?
- Upload evidenceGive the agent 12 months of statements and any billing emails; do not invent merchants or amounts.
- Review the listHave it list recurring items, ask which you still use, and flag duplicates, price rises and annual renewals.
- Rank cancel candidatesAsk for annualized savings, evidence lines and any exit fees before you decide.
- Approve then confirmApprove each cancel yourself. Afterward, have it confirm which payments actually stopped.
The prompt
The original prompt by @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, ChatGPT agent and Claude. Values in lime are placeholders.
Original prompt by @Michael_Fenech_
Review the last 12 months of statements I upload and billing emails you can access. List recurring subscriptions, their cost, next renewal and cancellation terms. Flag duplicate services, price increases and annual renewals coming up. Ask me which services I still use rather than assuming. Rank the cancellation candidates by annual saving. Show me the evidence and any fees. Get my approval before cancelling anything, then confirm which payments have actually stopped.
Exactly as published by @Michael_Fenech_ in X post: 4 money-saving audits (subscriptions, internet/phone, insurance, unwanted charges).
Our adapted version (optional)
Fill in the blanks (optional; the copied prompt uses your values)
Audit my subscriptions from evidence I control. Evidence: {statement_description}. Billing emails available: {billing_email_access}. Services I still use often: {still_use}. Must-keep: {must_keep}. 1) Build a table: merchant, amount, cadence, annualized cost, next renewal if known, cancellation terms if known, status (keep / unused / duplicate / price rise / annual soon / unclear), evidence note. 2) Ask me about any service you cannot classify from evidence alone. Do not assume I stopped using it. 3) Rank cancel candidates by annualized saving. For each, show evidence and any exit or notice fees. 4) Wait. Cancel only the merchants I approve, one by one. 5) After those cancels, re-check statements or inbox and confirm which payments actually stopped. Use only amounts and dates in the evidence. Mark gaps unclear. Never invent fees or renewal dates.
What can go wrong?
- Never let the agent cancel without a clear approval for that merchant.
- Annualize carefully; weekly and quarterly charges are easy to misread.
- Missing a statement month is not proof a subscription ended; mark gaps unclear.
- Self-reported savings in the source post are not a promise for your accounts.
Our take
Subscription creep hides in statement noise. Fenech's prompt is useful because it forces evidence first, asks which services you still use instead of guessing, ranks by annual saving and keeps a hard approval gate before any cancel. That last part matters: an agent that cancels alone can drop a tool you still need. We ship this next to Miles's monthly inventory because the source, depth and confirm-stopped step are different. The Miles routine is a light recurring pass with drafts you send; this audit is a year of statements plus billing emails, with ranked candidates, fee notes and a post-cancel confirmation that payments actually stopped. We do not repeat the post's $210 or $400 figures as facts; they are speaker-claimed, self-reported results. Keep card numbers out of places you do not control, and mark missing months unclear rather than assuming a subscription ended. In our dry run with a fictional year of statement lines and billing emails (two duplicates, one price rise, one annual renewal in 14 days, one unclear ATM-like merchant), the agent listed recurring items, asked which we still use, ranked cancel candidates with evidence and fee notes, refused to cancel without approval, and after mock approval confirmed which payments stopped versus which still appeared.
Where the time goes back: One pass over a year of statements surfaces forgotten charges and ranks them by annual impact before you spend time on support chats.
FAQ
How is this different from personal-subscription-audit-routine?
Will the agent cancel for me?
Are the dollar savings in the X post guaranteed?
Source and credit
The original prompt on this page is by @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: 4 money-saving audits (subscriptions, internet/phone, insurance, unwanted charges)
- Creator
- Michael Fenech (@Michael_Fenech_)
- Published on
- X
- Original date
- Oct 23, 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
- Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. Evidence-first list; ask-before-assume; approval gate; confirm-stopped step held.
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
- Oct 23, 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.