Check claimed bot completions for real
Add a supervisor bot that re-checks every claimed completion against the real files and history on a schedule, and pings you only when something does not match.
For owners running several bots on schedules who cannot personally verify every claimed done · Grok Bot · Intermediate
Researched, not run yetWhat that means
Copy the prompt
Fill in the [brackets] after you paste it.
You are my verification Bot. Your only job is checking claimed completions from my other bots.
Every [15 MINUTES OR HOUR], read the new lines in [FOLDER PATH]. Each line claims a finished task and names its proof: a file, a commit, or a link.
For each claim:
1. Open the proof. Confirm it is real and matches the claim: the file exists and changed, the commit is in the history, the link loads and shows the work.
2. If a claim does not check out, message me with what was claimed and what you actually found.
3. If everything checks out, stay quiet.
If I ask when you last ran, answer with the time and the claims you checked.
Never edit another bot's files. Never mark a claim verified without opening its proof. Never take over the task itself.Copied. Paste it into Grok Bot next.
Free PDF
Get this setup as a PDF
Enter your email and a print-ready copy of this setup opens here, to save as a PDF. The free starter pack opens too. Tap confirm in the email that follows to get occasional new setups. Unsubscribe anytime.
Already a subscriber? Unlock it on this device
Free PDF
A confirmation email is on its way too: tap confirm in it to get new setups by email.
Three ways to run this setup
Free
Copy and paste it yourself
Copy the prompt above into ChatGPT, Claude or Grok. It drafts. You approve.
Go to the prompt$9 once
Complete Pack
One payment, no membership. Opens the pack pages: Bot Safety Checklist, Second Brain Starter Vault, Voice File Starter, Tested Grok Bot prompt pack, Write it as code prompts.
See the Complete PackFree to start, VIP $19 a month
Save it as a bot in the guided setup app
The app turns this setup into a bot with the draft-only rules. 3 setups build free; the rest are part of VIP. A human reviews a prompt or setup for you within 24 hours, up to 2 a month.
Build it in the app
Want this one tested next? Vote for it.
What you get back
Illustrative. Your results will differ.
- MISMATCH 8:12 AM: daily-report bot claims the 7 AM brief posted. No new file in /reports since yesterday.
- YOU 8:27 AM: when did you last run? BOT: 8:15 AM, 6 claims checked, 6 verified.
- MISMATCH 2:40 PM: claim names commit a3f9c2; not in the repo history.
3 steps
- Copy the prompt.
- Paste it into Grok Bot.
- Fill in the [brackets], then read the answer before you send or use it.
Quick answer
Your bots share one cloud computer, so one supervisor Bot can read what all the others produced. Its only job: on a schedule, read each claimed completion from a shared folder, open the proof behind every claim, and ping you only on a mismatch. Silence alone proves neither that the check ran nor that everything matched; when it matters, use the routine's Test button or ask the supervisor when it last ran. One owner in the Grok Bot groups runs exactly this and caught a completion reference that did not exist.
How do you set it up?
Tools
- Grok Bot
- One shared folder on the bots' computer, such as /completions
- A supervisor Bot with one routine
Steps
- Make a completions folder on the shared computer. xAI's docs say files on the computer are visible to every Bot, which is what makes one supervisor possible.
- Give each working bot a standing line: when you finish a scheduled task, append one dated line to the folder naming what you finished and where the proof lives (file, commit, or link).
- Create the supervisor Bot. The docs recommend one clear job per Bot: this one's job is verification, never the work itself. Say so in its first message and ask it to remember.
- Add its routine: on a schedule, read the new claim lines, open each proof, and confirm it is real and matches the claim. The file exists and changed, the commit is in the history, the link loads and shows the work.
- Set the report rule: message you only when a claim fails. Silence alone proves neither that the check ran nor that everything matched, so keep a way to check the check: the routine's Test button, or ask the supervisor in chat when it last ran.
- Test before you rely on it: use the routine's Test button or one chat run on safe input, because the docs note a test run performs real work. When it catches a false claim, fix the working bot's instructions, not just the task.
Where to paste this prompt
Grok Bot → paste the prompt into a chat, or save it as a routine or scheduled task if the job repeats
More ways to run it
Which chat tools work with this prompt?
Paste the prompt into any of these assistants (free or paid tiers).
- Grok Bot
How sure is this?
Last updated: · Last reviewed:
Before you use this
What it is allowed to do: Read-only. Only reads, researches, or summarizes. It never drafts outbound content or changes anything.
At a glance: time, cost and difficulty
At a glance
All outcome figures are estimates from cited sources, not independent verification by us.
- What it is allowed to do
- Read-only
Only reads, researches, or summarizes. It never drafts outbound content or changes anything.
- Time saved (estimate)
- Not measured yet
- Difficulty
- Intermediate
- Tools needed
- Grok Bot
- Typical cost
- Prices: Not stated in sources.
Reviewed Nov 3, 2026
Basis (estimate): No timing claim: the sources describe routines and shared files, not minutes saved.
What usually goes wrong?
Where agents tend to stall on this job. More patterns in Where AI gets stuck.
- Letting the supervisor fix things. Verification and repair in one bot hides the failures you most need to see.
- Claims without proof locations. Reject the line format, not just the claim.
- Checking too rarely. A daily check on hourly work means a full day of unverified dones.
What tips make it smoother?
- Silence alone proves neither that the check ran nor that everything matched. When it matters, use the routine's Test button or ask the supervisor in chat when it last ran and what it checked.
- Keep the proof formats few: file, commit, link. Exotic proofs slow the check.
- A false claim is an instructions bug in the working bot. Fix the rule, then re-run.
Where do the claims come from?
Summaries use public sources linked below, not independent verification. How setups are picked.
- Read xAI’s write-up (docs.x.ai)
xAI: Create and manage Bots: give each Bot one clear job with approval boundaries, and set it in the first message with a request to remember (read 2026-11-03)
- Read xAI’s write-up (docs.x.ai)
xAI: Create a routine: plain-word schedules, the Test button, and the warning that a test run performs real work (read 2026-11-03)
- Read xAI’s write-up (docs.x.ai)
xAI: One computer, shared by all your Bots: files are visible to every Bot, so a supervisor can read the other bots' output (read 2026-11-03)
Common questions
- Why not just trust the bot's word?
- Because a claim is a chat message and the work is an artifact. Owners in the Grok Bot groups caught a claimed completion reference that did not exist; checking claims against files and history is what separates done from said-so.
- How often should the supervisor run?
- Often enough that a false claim cannot age: every 15 minutes to an hour for daily work. The docs support plain-word schedules, and a narrow read-only check is a light routine.
- Can the supervisor fix what it catches?
- Keep it read-only. A bot that verifies and repairs the same work grades its own homework, and you lose the signal.
Free kit by email
Get the free kit by email
The starter pack and the Portable AI Office, sent to your inbox. New setups by email after that. Unsubscribe any time.
What to do next
- Organize your bots like an org chart
Lay out your bots like an org chart: one coordinator, specialists with one job each, clear handoffs, and your approval on anything that goes out.
- Audit your bot instructions every week
Keep a fleet of bots honest about what they were told, with a weekly check that flags instructions that slipped or conflict.
- Catch a long agent run drifting off course
Add a written goal, regular goal checks, a step limit and a stop-for-review checkpoint to long agent runs, so drift shows up mid-run instead of at the end.
Have a setup that works? Submit it. We credit sources and write it up in plain language.