> ## Documentation Index
> Fetch the complete documentation index at: https://docs.oauth.fyi/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> What AEGIS protects your server against, and whether it's the right fit.

AEGIS is a Discord security platform, not a generic moderation bot. It runs on one operating
loop: **Observe → Detect → Contain → Investigate → Enforce → Audit → Improve.** Every step of
that loop is backed by durable state and an append-only, hash-chained audit log — nothing lives
only in memory, and nothing enforces without a record of why.

## Is AEGIS for you?

AEGIS is built for servers that need more than "ban the obvious troll." It's a good fit if:

* You've had — or worry about — coordinated raids, mass-ping floods, or invite spam.
* You want protection against a compromised staff account or a webhook being used to nuke
  channels and roles, not just protection against ordinary members.
* You need moderation actions to leave a real record: who did what, why, and whether it actually
  applied — even when Discord itself rejected it.
* You want an incident-response process for anything bigger than a single warn or ban.
* Your team is more than one person, and who can approve what actually matters to you.

If you just want a lightweight warn/ban command with no audit trail, configuration surface, or
detection engine behind it, AEGIS is more machinery than you need.

## What it protects against

<CardGroup cols={2}>
  <Card title="Raids" icon="people-group">
    Tracks join activity for coordinated bursts and escalates containment in stages instead of
    jumping straight to a lockdown.
  </Card>

  <Card title="Anti-nuke" icon="shield-halved">
    Per-actor action budgets catch mass channel/role destruction, webhook abuse, and unauthorized
    permission grants — including from a compromised staff account.
  </Card>

  <Card title="Malicious links & spam" icon="link-slash">
    Catches malicious/impersonating links, spam, mention floods, and invite spam with a
    progressive response, not an instant ban.
  </Card>
</CardGroup>

## What it gives your team

<CardGroup cols={2}>
  <Card title="Case management" icon="folder-open">
    Warnings, notes, and every moderation action land in a durable, sequential case record —
    even when Discord itself rejects the action.
  </Card>

  <Card title="Incident response" icon="triangle-exclamation">
    Raids, anti-nuke events, and other security incidents get a timeline, a commander, and
    optionally a private incident room.
  </Card>

  <Card title="Dashboard" icon="gauge">
    A web command center for cases, incidents, detection signals, audit search, and
    configuration — read access mirrors your AEGIS permissions.
  </Card>
</CardGroup>

## What it deliberately does not do

* No opaque ML risk scores — every score is a visible, deterministic list of factors, shown to
  moderators alongside the flag.
* No mass bans purely from join velocity — raids need corroborating signals, and lockdown
  defaults to a human approving it first.
* No identity correlation, doxxing, or tracking beyond what Discord already exposes.
* No claims of perfection: detection is tunable, and new protections start in a
  log-only or approval-required mode so your team can review before anything enforces.

## How it is built

**Detection explains itself.** Sentinel's risk scoring is deterministic, not a black-box model.
Every contributing factor is a plain-language reason, and the factor list — not just a number —
is what moderators see when they investigate a member or review an approval panel.

**Nothing escalates without permission.** Every protection has an enforcement mode, set per
guild, and new protections start cautious rather than enforcing immediately. See
[Configuration](/aegis/configuration) for how that works.

<Card title="Start here" icon="rocket" href="/aegis/quickstart">
  Add AEGIS to your server and run the first-time setup.
</Card>