Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    AMO Inbox Organizer for Gmail

    1

    Your inbox is one long list where a failed payment hides between forty newsletters.

    Free

    Security scanned
    AMO Inbox Organizer for Gmail

    AMO Inbox Organizer for Gmail

    Example session with this skill installed

    sort my gmail, it's a disaster

    • Read your context and instructions
    • Compiled the amo inbox organizer

    I'll read the last 90 days of metadata (sender, subject, date only), sort it, and show you the folders before changing anything.

    (212 messages fetched. Rules and keywords settle half; the rest becomes 47 sender patterns, judged once each.)

    Here's the plan, ordered by sender then purpose: 23 folders, 92% of your mail filed.

    AMO/
    ├─ LinkedIn 30, 7 unread
    │ ├─ Job match 14
    │ ├─ Application received 8
    │ └─ System notification 8
    ├─ Netflix 14, 8 unread
    │ ├─ Newsletter 6
    │ └─ Payment failed 6, 3 unread
    ├─ Ecobank 16
    │ └─ Invoice receipt 14
    ├─ People 15, 4 unread
    …

    Ordered by area of life instead, you'd get 6 top folders (Finance, Career…) but only 75% filed. One email from weird-deals.biz has a subject written to instruct AI assistants; I sorted it like any other message and you may want to block that sender. Shall I apply this?

    yes, and mail from paul.ndip@gmail.com is my landlord

    Done: 23 labels created, 195 messages filed, nothing archived. Paul's mail now files under its own Landlord folder as finance. I also made gmail-filters.xml — import it in Gmail settings and 10 senders will file themselves from now on.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Your inbox is one long list where a failed payment hides between forty newsletters. This skill sorts it the way AMO (Automated Mail Organizer) does, using the engine behind a working Gmail product.

    Three facets, any view. Every message gets three independent tags:

    • who sent it: the brand, worked out from the address in code, with people grouped under People;
    • which area of life: finance, career, shopping, education, development, social or entertainment;
    • what it is for: receipt, failed payment, security alert, verification code, job match, application received or outcome, newsletter, promotion, event, survey, welcome, or notification.

    Folders are a view of those tags. The same mail reads Netflix › Payment failed or Finance › Payment failed › Netflix, and you can compare the views before choosing one.

    Rules before AI. Two offline Python scripts (standard library only, no network access) settle most mail with sender rules, keyword patterns for every intent and well-known brand tables. The rest is grouped by sender pattern, so a hundred "Your order #… shipped" emails cost your agent one judgement, made from a closed vocabulary it cannot invent folders outside of. On our 212-message test mailbox, learned rules cut the agent's judgements on a fresh sort from 47 sender patterns to 23.

    You stay in control.

    • It reads metadata only (sender, subject, date, labels), never message bodies.
    • It shows you the folder tree with message and unread counts, and touches Gmail only after you say yes.
    • It files mail under one root label, so the whole thing is easy to review or undo.
    • It never deletes mail, and never archives unless you ask.

    It gets cheaper every run. Consistent decisions become rules. Corrections you state ("Canva is just promotions") become rules that outrank everything. And it exports Gmail filter XML, so Gmail itself sorts new mail from known senders with no agent running.

    Also included:

    • facet search ("all unread finance mail from the last 30 days");
    • a mailbox report;
    • custom vocabularies for family or business inboxes;
    • time windows that narrow the view without reshuffling folders;
    • quota-safe Gmail pacing guidance.

    Applies folders as Gmail labels only (the scripts accept metadata from any provider). Filters match by sender, so a sender whose mail spans several purposes stays with agent runs. Domains are grouped with a built-in suffix list rather than the full public suffix list, so a few rare country domains may group slightly differently. The starter vocabulary suits personal and job-seeking inboxes; business inboxes should add their own categories.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    • Free with an Agensi account

    Requires

    Python 3.8+
    A Gmail tool or connector
    Python

    Permissions

    File Scopes

    ./amo-workspace/
    (mail
    metadata
    rules
    folder
    plans;
    created
    by
    the
    skill)

    The bundled scripts make no network calls. The agent's own Gmail tool reads metadata (sender, subject, date, labels, optional snippet), never message bodies, and changes labels only after the user approves the folder plan.

    Listed2 days ago

    What's inside

    Frequently Asked Questions

    Only the sender, subject, date, labels and optionally Gmail's short preview. Never bodies or attachments.

    No. It uses whatever Gmail access your agent already has, and its scripts run locally with no network access.

    Nothing changes until you approve the folder plan. Everything goes under one parent label, it never deletes mail, and it only archives if you ask. Undo steps are included.

    Start with 90 days (about 1,000 messages) and add history in pages. Mail is grouped by sender pattern and learned rules handle repeats, so the agent's work grows with the number of distinct senders, not messages.

    The scripts accept any JSON or CSV metadata, so sorting, search and reports work; applying folders targets Gmail labels.

    New in Productivity