Remote Ritual Prompts That Actually Work in Team Tools

1. Auto triggering team rituals using Slack slash commands and delays

We had this daily check-in ritual—just one emoji from everyone, dropped into a #morning-vibes channel—except it always relied on someone remembering. So we rigged up a workaround using a Slack slash command and a Make scenario. And it worked, mostly. Until it didn’t.

The trick uses a custom slash command like /firemorning, which hits a webhook in Make. Make would wait 10 minutes (to avoid immediate spamming), then post a question prompt like “What’s something non-work you’re excited about today?” Plus a random emoji or GIF pulled from Giphy for extra dopamine.

The gotcha: Slack forcibly re-authenticates incoming webhook connections if the Make webhook URL changes even slightly, and it doesn’t throw an obvious error. Instead, nothing happens after the delay. The slash command says success, but the message never posts.

If you dig around in the Slack logs (admin panel → logs → slash commands), the request gets marked as “delivered.” Meanwhile, Make sees…nothing. The moment you copy the same webhook into a different scenario or rename it, the internal signature check dies silently. The fix? Reinstall the Slack app connection inside Make to refresh credentials. It doesn’t tell you this anywhere—just dies and stares at you.

2. Rotating async check-in questions using Notion database and Zapier

At one point, we used a Notion database for our question queue. It held discussion prompts like “What’s a small recent win you haven’t shared?” and “What’s your current browser tab chaos level?” Once a day, Zapier would pick a new one and post it in our team chat. And that’s where the insanity started.

Turns out Zapier’s Notion integration has a hard 100-item read limit per query—even if your view returns 300. We had 129 questions. Every time the automation ran, it pulled from the same first 100 rows. Randomness worked, technically, but we kept getting repeats within a few days. I thought it was bad random logic until I printed the data and saw row 101 could never be touched.

My Notes-on-a-whiteboard fix:

• Created a “used” checkbox property
• Added a view in Notion that filtered to unchecked + published questions
• Made Zapier pull from that view instead
• When posting a question, it updated Notion to check it off

The workaround survived maybe three weeks before one teammate duplicated the page and broke the underlying view filter—since Notion resets sorting and filters when duplicating databases. This reset the “used” checkbox for every row in the new copy. Random started repeating again. Fun times.

3. Getting daily rituals into Obsidian using local folders and cron

I wanted our async standup answers to land inside Obsidian where I write everything anyway. Emails? Slashed. Docs? Summarized. Partners’ odd 11pm Slack messages? Filed and buried. The idea was to push each day’s set of answers into a markdown folder tagged #team-checkin.

The Zapier to local file jump didn’t work cleanly—so I swapped in an Integromat (sorry, Make) workflow that takes the chat answers, formats them into markdown, and dumps them via an SSH write to a synced folder. That folder then got slurped by Obsidian’s sync when opened next.

The missing doc quirk? Timestamps in filenames. Obsidian doesn’t like colons in filenames on Windows (they’re invalid). But the Make module I used defaulted to UTC timestamps like 2024–04–11T08:00:00Z, and converted that to 2024-04-11T08:00:00Z.md — which completely failed to save the file on my Windows teammate’s machine under Obsidian Sync. Just silent failure, no error message anywhere.

Eventually I found out you need to strip or replace : with - in the Make formatting step. This worked:

{{formatDate(now; "YYYY-MM-DD-HH-mm-ss")}}

Took me 45 minutes and two cups of tea to pinpoint it. Once that was fixed, rituals now flow into a shared vault with a search pane pinned to just show the #team-checkin notes. It feels weirdly magical seeing answers pop in from teammates scattered across timezones.

4. Handling ghost participants in ritual flows triggered by Google Calendar

A monthly ritual I genuinely love is our unstructured “Work Weird” session. It shows up as a calendar event with a link to a shared Figma board, some random constraints (e.g. only use Comic Sans, draw with non-dominant hand), and gets filled with absolutely unproductive joy. It also gets ghosted. A lot.

We used Zapier to trigger an advance Slack ping one hour before the event, pulling participant emails from the Google Calendar invite and tagging them. But here’s the gotcha: if someone’s RSVP’d “yes” in Google Calendar, but they’ve since removed the event from their calendar (say they declined later, or unsubscribed), Zapier still returns their email from the original attendee list.

Even if their actual status is declined, unless you explicitly add a filter to check for response_status == accepted, you end up tagging people who’ve silently bailed. They get annoyed, even if they never tell you. I found out because a teammate forwarded me the thread after getting pinged while on vacation, and her out-of-office status was visible in the earlier Slack thread. Oops.

The fix was filtering by event.attendees[].response_status == “accepted” inside Code by Zapier (because filters don’t handle nested arrays well). Here’s the painfully manual workaround:

return inputData.attendees
  .filter(att => att.response_status === "accepted")
  .map(att => att.email)
  .join(", ");

Low-key hate how often Google Calendar treats removed-attendees as still attached metadata. Felt very haunted.

5. Using ChatGPT to reword stale rituals without sounding robotic

We got stuck in a rut with our Friday wrap-up ritual. Everyone posted the same things week after week: “This week I did X, next week I’ll do Y.” Wildly uninspired. So we tried prompt engineering our way out of it.

Enter ChatGPT… but only after making some serious tweaks. Plain prompts like “Rephrase this status update to be more fun and conversational” led to weird, overly quirky outputs. Think: “Howdy teammate! Buckle up for my wild week of Jira adventures!” Which, no. Just no.

The useful prompt ended up being:

You are a project manager summarizing a short status update to share in a casual chat setting. Keep it clear, mundane, and lightly informal. Keep formatting minimal.

That one prompt struck gold. We stored it inside a Notion page and wrapped it into a Make flow. People dropped in bullet lists, the flow piped those into OpenAI, then pasted the rendered blurb into Slack under their name. Way more engaging, and nobody had to invent personality from scratch.

The unexpected friction? Rate limits. If multiple teammates submitted at once, OpenAI’s API threw vague “Too Many Requests” errors. Happens faster than you think if everyone pastes in bullets between 4:55-5:00 PM on a Friday. Added a queueing delay using Make’s sleep function for staggered calls. Slowed it down, but at least it didn’t break.

6. Exploiting Loom share URLs to embed videos in daily roundup posts

One teammate loved filming little 45-second Looms recapping their day, usually with a cat somewhere nearby. So we tried surfacing those inside our end-of-day Slack roundup. We’d paste the Loom URL and let Slack do the auto-unfurl. Here’s where it got weirdly fragile.

Loom’s embed previews stop working if the video is set to team-visible only. Even if everyone in Slack is part of the Loom workspace. Unless the video is “anyone with link,” Slack won’t show the thumbnail at all—it just displays a blank message or raw URL. No error, no broken preview—just emptiness.

The Loom docs never once mention this. I had to dig around Loom’s support forums and found someone else buried in the same hole. Turns out Slack checks for open graph tags and falls back to nothing if it can’t fetch metadata because of restricted privacy settings.

We ended up adding a Slack tip next to the form where people drop links: “If you’re sharing Looms, set the video to Viewable By: Anyone With The Link.” People still forgot. So we built a small validator in our Make flow that checked if the Loom URL started with https://www.loom.com/share/ and then used an HTTP module to look for .

If missing, it pinged the user to adjust privacy. Better than silence.

7. Surprise resets in Notion automations that silently break filters

There was a day when our weekly discussion ritual just… dropped the wrong prompt. Not an error, just an old question from months ago. That’s when I realized something deeper had snapped inside Notion’s database views.

The core cause? When someone edited the database view filter—to add a new sort rule—it silently reset the previously applied status filter behind the scenes. The visible UI still showed the right filter condition, but Zapier pulled inconsistent rows unless you reloaded the view and saved again. There’s no warning of this in Notion’s automation story. Nothing says: “Hey, you broke this view and cached rules are now invalid.”

The worst part: the filter showed “Published is checked”, but Zapier saw both checked and unchecked rows. Very normal-looking… until 9am hits and the bot drops a half-drafted prompt from April into the team thread.

I started treating Notion database views as stateful glass objects—fragile, meaningless outside of visual inspection. Built a small validation script to test expected record counts before allowing the post trigger to run. If it sees more than one result, the post gets skipped. Teammates now get pinged to manually choose the week’s prompt instead.