Privileged Access Management Overview

Twingate Privileged Access Management

Twingate Privileged Access Management is a Layer 7 access control layer that verifies user and AI agent identity on every Kubernetes API call, SSH session, and web app request. It uses the Twingate Gateway, an open-source reverse proxy that you deploy in your own environment.

Privileged Access Management has three protocol components: Privileged Access for Kubernetes, Privileged Access for SSH, and Privileged Access for Web Apps.

Use Twingate Privileged Access Management to:

  • Give engineers and AI agents Kubernetes access without kubeconfig tokens or cloud IAM credentials, using native Kubernetes RBAC.
  • Replace SSH bastion hosts and distributed SSH keys with short-lived certificates and recorded sessions.
  • Add SSO and group-based authorization to internal web apps such as Grafana and Jenkins without writing an OIDC integration.
  • Audit every kubectl command and SSH session, with logs and recordings kept on your own infrastructure.

Key Benefits

  • Unified Identity Enforcement: Your users authenticate once through your existing IdP. That identity passes through to every Resource they access, from Kubernetes clusters to databases to internal APIs.
  • Dynamic Zero-Standing Access: Access is granted just-in-time, based on identity, device posture, location, and context. Permissions are automatically revoked when no longer needed.
  • Complete Visibility and Control: Every command, API call, and database query is tied to a specific user or agent with forensic-level detail and session replay capabilities.
  • Cost-Effective Scale: Deploy comprehensive privileged access controls across your entire environment without needing hardware appliances. No complex infrastructure requirements.

How Twingate Privileged Access Management works

Privileged Access Management introduces the Twingate Gateway, an open-source Layer 7 reverse proxy deployed within your environment. The Twingate Gateway enables identity propagation and session recording within those environments.

Before any request hits your protected environment, Twingate authenticates the user and enforces access policies with no static credentials or separate authentication tokens required.

Traffic flows from the user’s device to the protected service in this order:

Twingate Client → Connector → Twingate Gateway → Kubernetes API server, SSH server, or web app

The Gateway then passes the user’s identity to the upstream service in the form that service expects:

  • Kubernetes: The Gateway impersonates the authenticated user against the API server, so Kubernetes RBAC and audit logs see the real identity.
  • SSH: The Gateway signs in to the target server with a short-lived certificate issued by an SSH certificate authority.
  • Web apps: The Gateway injects a signed JWT or identity headers into every request. It overwrites identity headers sent by the client, so a forged identity header can’t reach your app through the Gateway.

Requirements

RequirementApplies toMinimum
Connector versionKubernetes and SSH Resources1.82.0
Kubernetes versionClusters set up with the Twingate Kubernetes Operator1.31
kubectl versionAI agent hosts1.26
Twingate Client for macOSSSH2026.85
Twingate Client for WindowsSSH2026.90
Twingate Client for LinuxSSH2025.342
Client platformKubernetes and SSHmacOS, Windows, or Linux. Android and iOS are not supported.

Getting started with Twingate Privileged Access Management

For more information about deploying and using Twingate Privileged Access Management, explore our protocol-specific documentation:

Frequently asked questions

Is the Twingate Gateway required?

Yes. Every Privileged Access Management Resource is bound to a Gateway, which enforces identity on each request and records sessions. A single Gateway can serve multiple Resources in the same Remote Network.

Does the Gateway replace the Connector?

No. The Twingate Client reaches the Gateway through a Connector, so each Remote Network with a Gateway also needs at least one Connector.

Where is the Gateway source code?

The Gateway is open source under the Mozilla Public License 2.0. The source code, Helm chart, and release binaries are on GitHub.

Do audit logs or session recordings leave my network?

No. Audit logs and session recordings stay on your own infrastructure and are never uploaded to Twingate. The Gateway writes them to stderr by default, and it can be configured to write them to a file instead. You are responsible for forwarding and retaining them, for example in a SIEM or object storage.

How do I replay a recorded session?

Recordings use the asciicast v2 format. Paste a recording into, or upload it to, the Twingate session player.

Last updated