Get started
WORK, ACCOUNTED FOR

Your bot says it’s done.
See what actually landed.

Claim, evidence, check, and your decision — on separate lines.

A receipt, openedFictional walkthrough
SCScout · Research Bot8m ago

Research 10 competitors

CLAIM

“10 competitors researched.”

Reported complete by the bot
EVIDENCE

competitor-map.csv + 10 source links

CHECKED

CSV format and 10 rows

Source links not checked

Illustrated checks on fictional data; no live sites checked.
NOT DONE

2 pricing pages unavailable

YOUR ACCEPTANCEPending your review

Try Needs work. That note is what the bot can read next.

ON EVERY RECEIPT

Four lines. None of them is the whole story.

  1. 01

    The claim

    What the bot reported.

  2. 02

    The evidence

    The file or link it attached.

  3. 03

    The check

    What this app actually ran.

  4. 04

    Your decision

    Accept it, or send it back.

THE MOMENT

You were not in the chat. The receipt was.

  1. 01

    It filed complete.

    Scout marked the research done.

  2. 02

    The gap stays visible.

    Two pricing pages never opened.

  3. 03

    Missing receipt

    Monday brief. Expected by 9:00. Nothing arrived.

    Schedules watch reports. They don’t run bots.
  4. 04

    Your note goes back.

    The bot can read it on the next cycle.

THREE STEPS

Connect once. Keep the review loop.

  1. 1

    Sign in

    The workspace is ready.

  2. 2

    Connect

    Grok Bot Web or Cursor.

  3. 3

    Enable a bot

    It files a receipt.

Grok Bot Web SupportedCursor SupportedGrok Bot Desktop Known issue

Desktop sign-in is unreliable. Use Grok Bot Web or Cursor.

Explore the complete review loop

A small record. A sharper review.

  • 4lines on every receipt
  • 0chats or routines read
  • 2clients we support
  • 1note back to the bot

Start with the whole review loop.

FREE

$0/ month

Experience the entire review loop.

  • 3 bots · 100 work runs / month
  • 7-day history
  • Claims, evidence and checks separated
  • Accept, needs work and MCP feedback
  • Expected receipt monitoring
PRO · LAUNCH PRICE

$6.90USD / month

Launch price — early users keep it while subscribed.

  • 20 bots · 2,000 work runs / month
  • 90-day history
  • Full review loop and expected receipts
  • CSV and JSON exports
  • Longer evidence and review history

Checkout awaiting operator configuration

A few useful answers.

What is Bot Receipts?

Bot Receipts is a proof and accountability layer for autonomous Grok Bots and agents. It helps operators of multiple bots review delivered work without digging through every conversation.

What is an AI agent receipt?

A receipt is a structured report that separates the bot’s claim, direct artifacts, actual checks, unfinished work and the human’s acceptance or rejection. A complete claim is not an accepted result.

How does Bot Receipts connect to Grok Bot?

Add the Bot Receipts HTTPS URL as a custom MCP server in Grok Bot Web and complete OAuth. Selected bots submit receipts through MCP. An account connection does not automatically make every bot report.

Does Bot Receipts verify bot work?

It checks CSV and JSON structure and configured row/column criteria, and can check a public GitHub PR when you request it. Saved links and reference IDs are not independently checked. These checks do not prove factual accuracy; you decide whether to accept the result.

What happens when I reject work?

Your Needs work note is available through MCP. The authorized bot can retrieve and acknowledge it, then submit a preserved revision. Bot Receipts does not send a bot message or start work automatically.

How are missing scheduled receipts detected?

You set an expected receipt schedule and grace period. Bot Receipts flags an overdue expected report when none arrives. Schedules monitor reporting; they do not execute bots.

Does it work with Cursor and Grok Bot Desktop?

Grok Bot Web and Cursor, in the browser and the app, support the custom MCP connection. Grok Bot Desktop is currently limited by OAuth callback/session handling; use Grok Bot Web or Cursor instead.

Can I try the full review loop for free?

Yes. Free includes 3 bots, 100 admitted work runs per month, 7-day history, evidence review, acceptance, feedback and expected-receipt monitoring. Pro launch pricing is $6.90 USD per month; early users keep it while subscribed. Billing is currently in test mode.

Limited product analytics help us understand connection and review drop-offs. We don’t collect receipt text, evidence contents or credentials for analytics.