Skip to content

Daily junk folder triage with a review pile

A daily pass over an old mailbox's junk folder puts known senders back in the inbox, deletes sure spam and leaves anything doubtful for you. The browser runs headless and the model only returns a verdict per message.

PersonalWorks with Claude Code · Claude Desktop with MCPCommunity example

What it does

  • Opens the junk folder and the inbox in a headless browser once a day
  • Checks each sender against your known list and a copy of your main mailbox
  • Asks the model for a verdict from a closed list, with no tools
  • Restores legitimate mail, deletes confident spam and counts the rest as to review
  • Sends a notification only when a run fails or the review pile grows

Your approval boundaries

Approval required

These actions always wait for your yes:

  • Deciding any doubtful mail
  • Unsubscribing from anything
  • Signing in again when the session expires

You decide what happens next.

More about this workflow

This setup is for busy people who want their own day to run a little smoother. It takes a repeat chore off your plate and hands back something you can check quickly.

First, it opens the junk folder and the inbox in a headless browser once a day. Next, it checks each sender against your known list and a copy of your main mailbox. Then it asks the model for a verdict from a closed list, with no tools. After that, it restores legitimate mail, deletes confident spam and counts the rest as to review. Finally, it sends a notification only when a run fails or the review pile grows. It works with your email, each on its own login with limited access.

You get the prep work done and waiting for you. The agent stops and asks for your yes before deciding any doubtful mail, unsubscribing from anything and signing in again when the session expires, and says plainly if a step cannot be undone.

“Every email is treated as hostile.”
TheBenBenJ TheBenBenJ on GitHub · Oct 30, 2026
Before you startEmail accessScoped permissionsClaude Code or Claude Desktop with MCP
Explore required tools and model recommendations

Tools and access

  • Gmail

    Handles the email part of this job. Grant read-only plus compose (drafts) scopes.

    API required: Yes. Needs a Google Cloud project with an OAuth app and the Gmail API turned on. The agent signs in through OAuth with only the scopes you pick.

Official links only. Not sponsored, no affiliate links.

Model recommendations

Best pick

Claude Sonnet 5.5

Fast and careful with your inbox and calendar.

$2 per million input tokens, $10 per million output tokens.

Anthropic docs as of October 11, 2026

Cheaper backup

Claude Haiku 5.5

Cheapest Claude option for daily sorting.

From $0.10 per million input tokens and from $0.50 per million output tokens.

Anthropic docs as of October 11, 2026

Our own picks. Not sponsored, no affiliate links.

Build your prompt

Add your details, then copy the prompt into your tool.

1Your details

Fill these in and the prompt updates as you type.

the old account

addresses or domains

for example 0.8

2Your prompt

4 of 4 still empty
Fill these in before you run it: [MAILBOX TO CLEAN: the old account] [KNOWN GOOD SENDERS: addresses or domains] [SPAM CONFIDENCE CUTOFF: for example 0.8] [RUN TIME] You are the Daily junk folder triage with a review pile agent. A daily pass over an old mailbox's junk folder puts known senders back in the inbox, deletes sure spam and leaves anything doubtful for you. The browser runs headless and the model only returns a verdict per message. Use only the details filled in above. If any [BRACKETED] item is still blank or unclear, stop and ask for it first. Do not guess. What you do on your own: - Opens the junk folder and the inbox in a headless browser once a day - Checks each sender against your known list and a copy of your main mailbox - Asks the model for a verdict from a closed list, with no tools - Restores legitimate mail, deletes confident spam and counts the rest as to review - Sends a notification only when a run fails or the review pile grows What you must never do without a yes from a person: - Deciding any doubtful mail - Unsubscribing from anything - Signing in again when the session expires Rules: - Before any step on the list above, stop and ask. Say in plain words what you want to do, who it affects, and whether it can be undone. - If something cannot be undone, say so first, in one short line. - Write drafts, not final actions. A person sends, pays, posts, or deletes. - Keep a short log of every action you took and every source you read. - If you are unsure, ask instead of guessing.
View run instructionsPaste into: Claude Code, Claude Desktop with MCP

Your answers stay in this browser only.

Run your workflow

Connect your tools with only the access this setup needs, then follow these steps.

Steps

Do these in order. Test on old data before it touches live work.

  1. 1Connect Gmail with its own login, not yours. Grant read-only plus compose (drafts) scopes.
  2. 2Paste the prompt where your harness takes instructions. See the note under the Copy button.
  3. 3Do a dry run on old or test data first. Check every draft and every stop point by hand before you let it near live work.
  4. 4Read its log after the first real run. Only widen access once a week of runs looks clean.

Safeguards and trust score

The score describes this example setup’s access, approval boundaries, ability to undo changes, and scope.

Access18 / 30

Can draft things that send or spend, so it sits close to risky actions.

Human approval30 / 30

Waits for a yes on: deciding any doubtful mail, unsubscribing from anything, signing in again when the session expires.

Undo and logs14 / 20

Proposes changes to live records, so the setup needs a change log and a way back.

Scope20 / 20

Runs on its own login across 1 tool. Fewer connections means a smaller blast radius.

How to raise it

  • Access: Take away its ability to send, post or spend. Let it only write drafts, and have a person press send or pay from their own account +6
  • Undo and logs: Stop it from editing live records. Have it write the change as a proposal you apply, and keep a log of every change with a way to roll it back +3

Do all of these to reach 91, up from 82.

Want help getting there? Get an AI Trust Audit from Pixeldesigns

Reviewed by Trustcase (example setup). How scoring works

Embed badge

Show this trust score on your site or README. It links back here.

Trustcase trust score badge
Markdown
[![Trustcase trust score for Daily junk folder triage with a review pile](https://trustcase.dev/badge/daily-junk-folder-triage-with-a-review-pile.svg)](https://trustcase.dev/use-cases/daily-junk-folder-triage-with-a-review-pile)
HTML
<a href="https://trustcase.dev/use-cases/daily-junk-folder-triage-with-a-review-pile"><img src="https://trustcase.dev/badge/daily-junk-folder-triage-with-a-review-pile.svg" alt="Trustcase trust score for Daily junk folder triage with a review pile" height="20"></a>

Other workflows that use the same tools

  • Follow up drafter

    Sales. Writes the follow up email after a sales call so nothing goes cold.

  • Pipeline cleanup

    Sales. Finds stale deals and tells you which ones to close or chase.

  • Meeting prep

    Sales. Gives you a short brief on everyone in tomorrow's sales meetings.