Security and Trust

Built for the most privileged system you run

An endpoint management platform can run code as SYSTEM or root on every device you own. That makes it one of the most valuable targets in your environment. Here is how Patchblox limits what an attacker, a mistake or an AI could do with it.

Read: Before you hand AI the keys to your endpoints and network
Security and TrustControls built into Patchblox
By design
  • Your data stays yoursPrivate Hosted sends Patchblox no endpoint data and no telemetry.
  • Your own certificate authorityEach customer gets its own root and issuing certificates. The root key isn't kept on the server.
  • Signed before it runsAgents verify the signature on every script and task before anything runs.
  • No standing write accessChanges in Intune and Azure use the signed-in admin's delegated permission.
  • Outbound onlyAgents connect out to their gateway. Nothing listens on your endpoints.
  • Scoped permissionsHundreds of discrete permissions, scoped by device group and by task type.
  • Secrets kept secretEncrypted to the target device's key and never written to disk or logs.
  • Read-only AIA read-only MCP server, with no setting that turns write access on.

Where your data lives

  • Private Hosted runs entirely on your infrastructure. Patchblox receives no endpoint data and no telemetry.
  • Patchblox Cloud gives every customer separate databases and separated services, with a dedicated option.
  • Patch Radar is built only from public catalog data, never from customer data.

Certificates and identity

  • Each customer gets its own root and issuing certificates. The root's private key isn't kept on the server; you hold the recovery file.
  • The issuing key is non-exportable, in the TPM where one exists. Code signing has its own issuing certificate.
  • Agents hold a TLS certificate plus a TPM-backed identity key where available.
  • Every component authenticates with certificates, tenant identity comes only from the certificate, and database connections use client certificates.

Signed content

  • Scripts and tasks are signed when they're created or edited, and agents verify the signature before anything runs.
  • Catalog content is signed by Patchblox. Third-party installers must carry a valid vendor signature, checked at download and again at install.
  • The Patchblox CDN hosts only Patchblox's own content, never third-party installers.

Least privilege

  • Patchblox services run as virtual service accounts without admin rights.
  • No standing write access to Intune or Azure. Changes there, including policy restores, use the signed-in admin's delegated permission.
  • The connector app in your tenant has read-only application permissions.

Network

  • Agents connect outbound to their gateway. Nothing listens on your endpoints.
  • Gateways and repositories are designed to sit behind your load balancer or WAF.
  • TLS everywhere. Patchblox never listens on plain HTTP.
  • IP restrictions at gateways, repositories and the management server.

Access and audit

  • Role-based access with hundreds of discrete permissions, scoped by device group and by task type: someone can reboot servers without being able to run scripts.
  • Owners and access lists on reports, scripts and workflows.
  • Sign-ins and actions are recorded in the audit trail.

Secrets

  • Each secret is encrypted to the target device's key.
  • Secrets are never written to disk, logs, command lines or environment variables.

AI

  • A read-only MCP server for AI you run. There is no setting that turns write access on.
  • AI clients authenticate as their own service identity with short-lived tokens and read through a dedicated, restricted database role. Every query is recorded in the audit trail.
  • Patchblox sends no environment data to any AI provider.

Reporting a vulnerability

Email security@patchblox.com. Our /.well-known/security.txt lists the same contact, and we'll publish external security reviews and penetration test summaries here once they're complete.