300+ MCP tools across ZPA, ZIA, ZDX, ZCC, ZTW, ZIdentity, EASM, Z-Insights, and ZMS. Read-only by default, HMAC-confirmed writes, OneAPI authentication baked in.
Production-ready MCP server purpose-built for Zero Trust operations — drop it into any MCP-capable client and your agent gets the full Zero Trust Exchange as native tools.
Let AI agents triage threats, audit policies, and automate routine changes across the entire Zscaler Zero Trust Exchange — without leaving your assistant.
Quickstart300+ tools across ZPA, ZIA, ZDX, ZCC, ZTW, ZIdentity, EASM, Z-Insights, and ZMS — every major service of the Zero Trust Exchange, behind one MCP interface.
Browse servicesRun locally via CLI or Docker, or deploy to Amazon Bedrock AgentCore, Azure Container Apps / AKS / Foundry, or Google Cloud Run / GKE / Vertex AI Agent Engine.
Deployment optionsRead-only by default. Write tools are opt-in and individually allowlistable. Destructive actions require HMAC-confirmed elicitation tokens — prompt-injection-proof.
Security modelA single set of ZIdentity OAuth credentials authenticates the server to every Zscaler service — no per-product API key juggling.
Authentication guideFirst-class plugins for Claude Desktop, Claude Code, Cursor, Gemini CLI, Kiro IDE, and VS Code + Copilot — install in one command, start prompting.
Editor integrationsFrom zero to AI-driven Zscaler operations in under five minutes.
Create an API client in the Zidentity console — a single set of credentials (client ID, client secret, vanity domain) authenticates the server to every Zscaler service.
Authentication guideInstall with uv, pip, or Docker. Configure with a .env file and choose your transport (stdio, SSE, or streamable-HTTP).
InstallationDrop a single MCP config snippet into Claude Desktop, Cursor, Gemini CLI, VS Code Copilot, or Kiro IDE. The server appears as a tool your agent can call.
Editor integrationsTry "List my ZPA application segments", "Show the last 10 ZIA URL filtering rules", or "Run a ZDX deep trace from my San Francisco office".