PodQuery is an open-source clinical audit extension that connects your Glooko and Omnipod 5 diabetes data directly to Claude Desktop. Built around the Model Context Protocol Bundle (.mcpb) standard, it installs via a single double-click with zero terminal configuration, Docker containers, or local database setups required.
Rather than copying and pasting sensitive logs into a web browser, PodQuery operates as a "local-first" engine on your machine. It enables you to ask Claude natural-language questions about your glucose trends, time-in-range (TIR), basal/bolus balance, and algorithm behavior, returning both analytical breakdowns and interactive, clinical-grade browser charts.
What It Does
PodQuery exposes your historical diabetes records as a structured suite of analytical tools that Claude calls on demand during your conversation:
-
Summaries & Trend Analytics: Instant calculation of time-in-range, Glycemic Measurement Index (GMI), coefficient of variation, best/worst days, and hourly patterns over custom windows.
-
Granular CGM & Bolus Inspection: Full-fidelity 5-minute continuous glucose monitoring data paired with active pump parameters (ISF, carb ratios, targets) to evaluate dose efficacy.
-
Omnipod 5 Behavior Tracking: Visibility into closed-loop algorithm states, including automated suspensions, maximum delivery caps, and signal-loss events.
-
Interactive Browser Visualizations: Generates self-contained, color-coded HTML glucose reports complete with hoverable bolus markers and per-day filter chips directly in your default web browser.
-
"Tough Love" Endocrinologist Persona: An optional, selectable prompt that instructs Claude to deliver uncompromising, direct feedback on bolus timing, over-corrections, and basal drifts.