Oliver Kieselbach
729ee56bde
The 2026-09-04 audit found no bypass of the GA boundary, but a failed attempt left only a Warning trace without identity, and a non-GA tools/call for a GA-only MCP tool never reached the backend at all. This makes the boundary observable without adding a second alarm writer (D-185): Backend - PrivilegedRouteDenied (Security): raised from the policy middleware's deny path for every 403 by an authenticated caller on a GlobalAdminOnly route. Critical for a caller without a platform role, Warning for a Global Reader. Details carry upn, oid, tid, client source, MCP tool name, correlation id, reason and policy. Throttled one hour per caller+path per instance (PrivilegedDenialReporter, IPrivilegedDenialReporter seam). PolicyResult now surfaces the denied tier; the DENIED log line gains oid, tid, clientSource and mcpTool. - GET global/raw/access-probe: no-op GlobalAdminOnly route whose denial IS the signal (AccessProbeResponse DTO, catalog entry, parity test). - Defense in depth: AppInsightsQueryFunction and both TableQueryFunction triggers re-check IsGlobalAdmin in the body (RawGlobalAdminGate); an empty RequestContext fails closed. - Dual-register: OpsEventTypes.All, OpsAlertRulesSection.tsx, regenerated manifests. MCP - ga-tool-probe.ts: before dispatch, a caller without the GlobalAdmin role naming a strictGa tool leaves an unconditional [mcp-security] line (tool, upn, scope) and fires the access probe with X-MCP-Tool-Name; the SDK answer stays "Tool not found". - GA_STRICT_TOOL_NAMES exported and pinned equal to the registered GA-minus-Reader set. - debug-session prompt names query_backend_logs/query_table only for strictGa (Readers were told to call tools they cannot see); tool_call_rejected lines carry scope + upn; config-tool comment corrected against the actual catalog tiers. Tests: reporter (severity, details, throttle, fail-soft), middleware policy surfacing and IsPrivilegedDenial/BuildPrivilegedDenial, Program.cs middleware order source scan, raw gate 403, probe catalog + JWT pins, wire parity; MCP probe unit tests, Global Reader over HTTP on both eras, runtime refusal fires the probe, prompt gate. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Autopilot Monitor
Advanced monitoring and troubleshooting solution for Windows Autopilot deployments. Gain full visibility into every enrollment session with a detailed event timeline, fleet health dashboards, and session reporting. Define custom analysis rules to automatically detect issues and gather rules to collect targeted evidence. Retrieve diagnostics packages on demand, configure agent settings like auto-reboot behavior and automatic timezone adjustment — all managed centrally from the web dashboard.
Availability
Autopilot Monitor is publicly available — the Community plan is free. Sign in with your work account at autopilotmonitor.com; new organizations complete a short activation step after first sign-in.
Overview
Autopilot Monitor provides real-time tracking, intelligent diagnostics, and automated troubleshooting for Windows Autopilot enrollment processes. It consists of:
- Bootstrap Script — PowerShell script deployed via Intune that starts monitoring early in the enrollment process
- Monitoring Agent — Lightweight .NET application that collects telemetry and evidence during enrollment
- Backend API — Azure Functions-based ingestion and processing pipeline
- Web Dashboard — Next.js application for real-time monitoring and fleet analytics
- MCP Server — Model Context Protocol server that lets AI assistants query and troubleshoot enrollments in natural language
AI Integration (MCP)
Autopilot Monitor ships a hosted Model Context Protocol server. Connect Claude Desktop, VS Code, or any MCP client that speaks Streamable HTTP with OAuth, and ask questions like:
- "Show me all failed enrollments from the last 24 hours and group them by failure reason."
- "Why did session X fail, and which app install caused the delay?"
- "Which devices in my fleet are affected by CVE-2024-30078?"
Server URL: https://mcp.autopilotmonitor.com/mcp — sign-in runs through your existing work account, access is scoped to your tenant exactly like in the portal, and no credentials are stored. Setup guide: AI Integration (MCP).
Documentation
Full admin documentation is available at docs.autopilotmonitor.com
Feedback
Bug reports and feature requests go through Issues, questions and ideas through Discussions. Security issues are reported privately, see SECURITY.md. For a problem in your own tenant, the Report Session button in the portal is the fastest path.
License
This project uses a split licensing model. The root LICENSE is AGPL-3.0, which covers the server-side components; the device-side components carry their own MIT license file:
- MIT License — Agent (
src/Agent/) and Shared library (src/Shared/) — unrestricted use on end-user devices, no copyleft obligations - AGPL-3.0 — Backend (
src/Backend/), Web Dashboard (src/Web/), and MCP Server (src/McpServer/) — modifications to server-side components remain open source, especially when deployed as a network service
The Shared library is MIT because it is a dependency of the MIT-licensed Agent. The license file inside each component directory is authoritative for that component.