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

# Hydra

> Hydra is the integration specialist among the Myrmex AI agents — it operates your connected tools across four domains: network devices, security software, databases, and cloud platforms.

import { CardGroup, Card, Note, Tip } from '@mintlify/components';

**Hydra** is the integration specialist among the Myrmex AI agents. It reaches beyond your endpoints to operate the tools you've [connected to Myrmex](/documentation/integrations/overview) — running queries and actions on firewalls, cloud accounts, SIEMs, databases, and more, so the whole environment responds to a single request.

## Four Domains

Hydra is one agent with four areas of expertise. Together they form the backbone of the [integrations catalog](/documentation/integrations/overview):

<CardGroup cols={2}>
  <Card title="Network devices" icon="network-wired">
    Switches, routers, firewalls, and other network gear — operated over SSH and
    their native protocols.
  </Card>

  <Card title="Security software" icon="shield-halved">
    SIEM, EDR/XDR, vulnerability management, threat intelligence, and ITSM
    platforms — queried and acted on through their integrations.
  </Card>

  <Card title="Databases" icon="database">
    Relational and other data stores — for health checks, configuration audits,
    and read-only inspection.
  </Card>

  <Card title="Cloud & platform" icon="cloud">
    Cloud providers and platform services — inspected and configured through their
    APIs and command-line tooling.
  </Card>
</CardGroup>

## How Hydra Operates Your Tools

You don't hand Hydra credentials in chat. Each integration is set up once — with its credentials stored securely by the platform — and Hydra then acts through **integration actions**: it references the integration and the operation to run, and the platform resolves the credentials server-side. Your secrets never pass through the AI.

Depending on the integration, Hydra reaches the target through a [Collector](/documentation/agent-endpoint-security/collector-mode) running on your network, over an SSH session, or via the platform's own connectors — but from your side it's the same experience: ask, review, approve.

<Note>
  Hydra is an **AI agent**. The **software agent** installed on a host is what
  can run as a [Collector](/documentation/agent-endpoint-security/collector-mode)
  to bridge Hydra to systems inside your network. Same word, different things.
</Note>

## AI Proposes, You Approve

Reading and querying your integrations happens freely, but anything Hydra does that **changes** a connected system — a firewall rule, a cloud setting, a ticket — is proposed for your review first. You approve before it runs.

<Note>
  Hydra's security domain overlaps with [Orion](/documentation/multi-agent-system/orion)
  for threat intelligence and reconnaissance. Hydra operates the tools you've
  connected; Orion focuses on investigating threats and indicators.
</Note>

## Where to Go Next

<CardGroup cols={2}>
  <Card title="Integrations overview" icon="plug" href="/documentation/integrations/overview">
    Browse the catalog and set up the tools Hydra operates.
  </Card>

  <Card title="Collector mode" icon="tower-broadcast" href="/documentation/agent-endpoint-security/collector-mode">
    How the software agent bridges Hydra to systems inside your network.
  </Card>

  <Card title="Centurion" icon="brain-circuit" href="/documentation/multi-agent-system/agent-orchestrator">
    The orchestrator that brings Hydra in.
  </Card>

  <Card title="Orion" icon="shield" href="/documentation/multi-agent-system/orion">
    The threat intelligence analyst Hydra works alongside.
  </Card>
</CardGroup>

<Tip>
  Scope a request to one system by `@`-mentioning the integration — Hydra then
  acts on just that tool.
</Tip>