Exploring the Codebase
Get oriented fast — an AI-generated codebase wiki, plus the authoritative architecture docs.
GlycemicGPT is a multi-component monorepo (backend API, web app, AI sidecar, Android phone + Wear OS apps, device-data plugins). This page is a starting point for finding your way around.
Ask DeepWiki (AI-generated overview)
DeepWiki is an auto-generated, AI-powered wiki of this repository. It maps the architecture, summarizes components, and lets you ask questions about the code in natural language — a fast way to build a mental model before reading source.
It refreshes automatically (the DeepWiki badge in the project README keeps it on a weekly re-index), so it tracks the default branch without manual upkeep.
It is AI-generated, and can be incomplete or wrong. Treat it as an orientation aid, never an authority — especially for anything safety-related. The source code, this
docs/site, and the safety rules in CONTRIBUTING are the source of truth. When DeepWiki and the code disagree, the code is right.
The authoritative references
When you need the real picture, start here:
- Plugin Architecture — how device-data drivers (pumps, CGMs) plug in, and the safety-limit contract every plugin must honor.
- Mock Data Service: how the web app simulates CGM, pump, Nightscout, cloud sync, and live stream API data during development.
- Wear OS Architecture — the phone ↔ watch data layer.
- Branching Strategy — the develop → main flow and how releases are cut.
- Local Dev Testing Checklist and Security Testing — what to run before opening a PR.
Repository layout (top level)
apps/api— FastAPI backend (Python).apps/web— Next.js web app.sidecar/— the AI sidecar (provider abstraction for BYOAI).apps/Android modules +plugins/— the mobile app, Wear OS, and device-data drivers.docs/— this documentation site.