Zscaler AIRT Manager MCP Server

An MCP (Model Context Protocol) server for interacting directly with the Zscaler AI Red Teaming (AIRT) SaaS platform. This server allows AI agents to programmatically manage targets, query active campaign test runs, extract detailed probe states, trigger new vulnerability sweeps, and trigger native platform retries for error healing.

Capabilities

• list_ai_apps: List all configured chatbot targets and LLM applications.

• get_ai_app: Fetch detailed configurations for a specific AI Application.

• get_app_probe_settings: Fetch active/inactive probe template configurations for an app.

• trigger_test_run: Trigger a new Campaign Test Run (customizable probe IDs or full sweeps).

• get_test_run_status: Fetch the master progress, pass/error counts, and execution state of a run.

• get_test_run_probes: Drill down into the specific completion and error metrics of individual probes within a run.

• continue_probe_run: Trigger the native Zscaler platform /continue action to intelligently retry failed (e.g. timeout) test cases inside a locked run.

• cancel_test_run: Abort an active or hanging Campaign Test Run.

Related MCP server: mcp-defectdojo

Setup

1. Make sure you have uv installed (pip install uv).

2. Run uv sync to install dependencies.

3. Configure your environment variables in .env (see .env.example).

Environment Variables

• ZID_CLIENT_ID: Your Zscaler ZIdentity Client ID.

• ZID_CLIENT_SECRET: Your Zscaler ZIdentity Client Secret.

• ZID_TOKEN_URL: The ZIdentity OAuth2 token endpoint (default: https://hshendemo.zslogin.net/oauth2/v1/token).

• ZSCALER_BASE_URL: The Central Plane API Base URL (default: https://api.zsapi.net).

Usage with FastMCP

You can run this server using the FastMCP CLI or include it in your Claude Desktop / Gemini CLI config.

fastmcp run server.py