Inbox Rule Abuse Triage logo

Inbox Rule Abuse Triage

Organization
CDESpace
Inbox Rule Abuse Triage

Triage an alert where a mailbox rule was created or changed in a way that hides mail — auto-forwarding to an external address, or moving and marking read anything matching finance, security, or password keywords. Reads the audit trail and the rule definition to tell an organizing user or a helpdesk fix apart from a rule planted to intercept replies or bury alerts after a mailbox takeover, and decides escalate or dismiss.

Overview

PublisherCDESpace
RepositoryDetection-Skills
Skill nameInbox Rule Abuse Triage
Stars
57
Forks
5
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by CDESpace on GitHub. Read the source before you install it.

Installation

Install the Inbox Rule Abuse Triage AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/CDESpace/Detection-Skills/tree/main/skills/triage/Inbox%20Rule%20Abuse%20Triage
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Use it in TypingMind

Enable Inbox Rule Abuse Triage in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Inbox Rule Abuse Triage on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Inbox Rule Abuse Triage is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Triage Steps

This class fires on a mailbox rule change — a new or edited rule in Exchange Online, Google Workspace, or another mail platform. Most rule changes are housekeeping: people file newsletters into folders, forward their mail to a personal account before a holiday, mark a mailing list as read. A rule is also one of the quietest footholds an attacker keeps after a mailbox takeover, because it survives a password reset and needs no further sign-in: it silently forwards replies to a hijacked thread, or deletes the security notifications that would give the intrusion away. The signal is not that a rule changed — it is what the rule does and whether the account it changed on is otherwise clean.

Begin with the state of the mailbox's owner: a rule created minutes after a sign-in that triage already flagged, or on an account with an open takeover alert, starts the call near escalate. Then read the rule definition and the audit event for the leads below. They stack; a single strong one — forwarding to an external address the user never mailed, or a rule that deletes anything saying "security alert" — can carry the call.

Leads that point to a real threat — what to look for in the data:

  • Forwarding out of the org. A rule that forwards or redirects mail to an external, personal, or newly-registered domain — the classic reply-interception and data-siphon shape. In the audit log: a New-InboxRule / Set-InboxRule (or the Google equivalent) with ForwardTo, RedirectTo, or a forwarding SMTP address outside the tenant. It weighs most when the destination is a domain the user has no mail history with, or when org-wide auto-forwarding is supposed to be disabled and this rule is an exception.
  • A rule that hides, not organizes. A rule keyed on words an attacker wants the user not to see — invoice, payment, wire, password, security alert, sign-in, MFA, the attacker's own display name — that moves matches to Deleted Items, RSS, Archive, or Conversation History and marks them read. Deleting or hiding is the tell; foldering into a clearly-named folder the user reads is not.
  • Created in a takeover window. The rule change lands in the same session or hour as a sign-in that triage flagged, an MFA registration the user never started, or a token replay — persistence set immediately after access. The rule and the suspicious login sharing a source IP or client is the strong version.
  • Set by someone other than the owner. The audit event shows the rule created by a different principal, a delegate, or an OAuth app rather than the mailbox owner — especially an app with mail-write scope the user doesn't recognize.

Leads that rule it out — benign context you can actually see in the data; if the data doesn't show it, you don't have it:

  • It organizes, and stays inside. The rule files matches into a named folder the user opens, or forwards to another internal mailbox for a shared workflow — no external destination, no delete, no mark-as-read on security mail.
  • It fits the owner and the moment. The rule was created by the mailbox owner from a device, network, and session that positively match their baseline, with no takeover signal anywhere near it — an out-of-office forward to a teammate, a filter for a project alias.
  • A helpdesk or admin change on record. The change traces to a documented migration, a delegation the user requested, or an admin acting on a ticket.
  • A known automation. A mail-integration app the tenant sanctioned, creating the rule on its normal schedule with scopes the org approved.

To confirm a lead instead of guessing, pull the thread: after the rule appeared, did mail matching it actually get forwarded or deleted, and did the same session touch a sign-in, an MFA registration, or a file download? A rule plus activity in the same window is a confirmed foothold, not a suspicion.

Output

Decision

  • escalate: a rule that forwards or redirects to an external or unfamiliar destination, or that deletes/hides mail matching finance or security keywords — especially one created in the same window as a flagged sign-in or MFA change, or set by a principal other than the owner. Any rule paired with a takeover signal on the same account escalates on its own.
  • dismiss: a rule that files mail into a folder the user reads or forwards internally for a known workflow, created by the owner from a session that matches their baseline with no takeover signal near it, or a change traced to a documented helpdesk or admin action. A dismiss is a positive call that the rule is benign, made with the rule definition and the owner's context in hand — not the mere absence of a forward. When the destination, the intent of the rule, or the owner's context is unclear, escalate.

Evidence

The rule definition (conditions, actions, destination address), the audit event and who created it, the mailbox owner's takeover status and sign-in baseline around the change, whether matching mail was actually forwarded or deleted, and any sign-in, MFA, or OAuth activity sharing the session.

Reasoning

Name the leads that decided it and how they stacked — a rule forwarding invoices to a personal domain, created minutes after a sign-in from a new ASN, is escalate on its own; an owner-created filter moving a newsletter to a named folder from their usual device is dismiss.

Frequently asked questions

What does the Inbox Rule Abuse Triage AI skill do?

Triage an alert where a mailbox rule was created or changed in a way that hides mail — auto-forwarding to an external address, or moving and marking read anything matching finance, security, or password keywords. Reads the audit trail and the rule definition to tell an organizing user or a helpdesk fix apart from a rule planted to intercept replies or bury alerts after a mailbox takeover, and decides escalate or dismiss.

Why use Inbox Rule Abuse Triage on TypingMind?

Because you install it once and use it with any model. Inbox Rule Abuse Triage is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Inbox Rule Abuse Triage in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/CDESpace/Detection-Skills/tree/main/skills/triage/Inbox%20Rule%20Abuse%20Triage. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Inbox Rule Abuse Triage?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Inbox Rule Abuse Triage?

As many as you like. As long as a model supports skills, you can use Inbox Rule Abuse Triage with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Inbox Rule Abuse Triage AI skill free?

Yes. It is published on GitHub by CDESpace under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇