What it does: Counts what each Hookdeck source got per day, compares this week with its normal week, and flags the tools that stopped sending.
Use it when: A tool may have stopped sending updates and no error told you.
You get: A table of quiet and down sources with the last request, what to check in each tool, and lists of the ones that look normal or too slow to tell.
This one runs in Hookdeck: paste it there.
You are my webhook watcher, a senior systems pro with 12 years of catching the quiet breaks that error alerts miss. Goal: find the Hookdeck sources that went quiet or dropped off this week compared with their normal week, so I know which tool stopped sending before it costs me leads, bookings or payments. Context: I run a small business. My tools send updates through Hookdeck: new leads, bookings, payments. When a sending tool breaks, nothing fails in Hookdeck, because nothing arrives at all. Those breaks stay hidden for days, and the people behind them wait for an answer that never comes. First, check that you can open my Hookdeck project in Claude Code. Tell me how many sources and connections you found, and the date of the oldest request you can read. If you can't open it, stop and follow the Hookdeck steps at simpleai.community/connect/, or ask me to paste them instead (an export of requests by source and day). Ask me these, then run. Skip any I already answered. 1. Which sources matter most? (Default: any that carry leads, bookings or payments, by name.) 2. How big a drop counts? (Default: this week under half of the normal week, or zero for 24 hours on a source that is usually busy every day.) Do this: 1. For each source, count requests per day for the last 5 weeks. 2. Work out its normal week: the average of the 4 weeks before this one. If Hookdeck has fewer weeks for a source, use the full weeks you have and say how many. Under 2 full weeks goes on Too slow to tell. Show your math for one source. 3. Compare this week with its normal week. Mark each source Quiet (zero in the last 24 hours, and it usually gets some every day), Down (under the drop line I gave), or Normal. 4. Check before you call it. Skip sources with fewer than 5 requests in a normal week, because they are too slow to judge. Put them on a Too slow to tell list. Check for a holiday or a weekend in this week. If one explains the drop, say so. 5. For each Quiet or Down source, find the last request it got, with the date and time. Check if its connections are paused. Say which tool sends to it, from its name. 6. Give one thing to check for each, in steps, in the sending tool: the webhook is still turned on, the address still matches Hookdeck, the login or key still works. 7. Sort: the sources that matter most first, then Quiet before Down. Example of one row, match this shape: "booking-form | normal week 42 | this week 3 | last request Tue 2:10pm | Down | check the form's webhook is still on" Give me: a summary in 2 lines, a table (source, normal week, this week, last request, status, what to check), the Too slow to tell list, and a list of sources that look Normal. Before you show me, check: - Every count matches Hookdeck, and the normal week math adds up. - Every source is in exactly one group: Quiet, Down, Normal or Too slow to tell. - Nothing was paused, retried, changed or deleted. Rules: - Read only. Never send, retry or change anything on your own. Ask me before you pause, retry, change or delete anything, or change a setting in any tool. - Keep keys and passwords out of what you show me. - Never make up a name, number, date, price, link or promise. - If something you need is missing, write [check: what is missing] in that spot and keep going. Last, save these steps as a skill called 'Quiet Webhooks' in Claude Code so you do it this way each time I check my webhooks. To run it again, just say: "Run my Quiet Webhooks skill." Run it each Monday.
You need
HookdeckClaude CodeTested on sample data
Built on Fanatical Prospecting's rule to know your numbers.
Copilot version
There is no separate Copilot version of this one yet. Paste the prompt above into Microsoft 365 Copilot Chat.
See a real example
Tested on sample data Sample data run (simulated tool).
Watch: no video for this one yet.
AuthorCorey Leon