"Which of my endpoints are missing critical patches?" Ask your AI assistant, and Automox MCP Server pulls the answer live from your console. When you ask it to fix them with a Patch by Severity policy, it drafts the policy, and you can review the blast radius before anything runs. Nothing changes until you approve it.
Most AI integrations cover a slice of a product. Automox MCP Server covers the full published Automox Console and Webhooks APIs through the Model Context Protocol, the open standard assistants use to work with outside systems. Endpoints, policies, patches, groups, webhooks, Automox Worklets™, vulnerability sync, maintenance windows, and reporting are all reachable in plain language. The only operations left out are the ones that would expose a secret.
Coverage that wide only works if you trust it. Run it in read-only mode and exploring your environment can't change anything. It asks for confirmation before every write, and it keeps high-impact operations locked until you opt in.
The 2.2 update is mostly about what you can see and build without leaving the assistant.
Interactive review surfaces. In assistants that support them, patch approval queues, compliance triage, a policy's blast radius, remediation actions, and role-based access control (RBAC) certifications render as something you can read and click through, not paragraphs to sort out.
Patch by Severity policy creation. Describe the severity levels you want covered, and the assistant builds the policy, instead of you assembling it by hand in the console.
Live capability discovery. The assistant sees which tools it can use in the current session, which are held back, and which are safety-gated. When a tool is safety-gated, it names the exact setting you need to change.
Structured fallback. Assistants that can't draw the visual surfaces receive the same details as structured data, so no one loses a capability because of their client.
Start in read-only mode. Explore your posture, compliance, and patch readiness with nothing at risk, then turn on write access when you're ready.
The MCP Server 2.2 overview lays out everything in the release, and the Automox MCP Server page has setup instructions.
Frequently asked questions
It's a Model Context Protocol server that connects an AI assistant to the full published Automox Console and Webhooks APIs, leaving out only operations that would expose a secret. You run endpoint operations in natural language through one governed interface, instead of a narrow connector that reaches only a few workflows.
You can run Automox MCP Server in read-only mode. It asks for confirmation before any write action and gates high-impact operations behind explicit opt-in. Automox excludes secret-exposing operations entirely, and the assistant never assumes access it wasn't given.
Version 2.2 adds interactive review surfaces, Patch by Severity policy creation from natural language, live capability discovery that shows which tools are available or safety-gated, and structured fallback for hosts that can't render visual surfaces.
It lets the assistant see exactly which tools are available, withheld, or safety-gated in the current session. When a tool is safety-gated, the assistant surfaces the specific setting you need to change to unlock it.
Start in read-only mode to explore endpoint posture, compliance data, and patch readiness with zero write risk, then enable write workflows at your own pace. The Automox MCP Server page walks you through setup.
)
)
)
)