Internals

The deep technical notes, kept as they were written. They document why the code is shaped the way it is; the topic sections document what it does.

Note	About
Architecture	System overview, database schema, multi-worker startup, vector search, auth, rate limiting, concurrent session protection
Services	The backend service layer — AI extraction, embeddings, vector search, PDF processing, email, reminders
Frontend	React structure, pages, components, state patterns, runtime configuration
Deployment	Docker, environment, HTTPS, rebuilding, monitoring, scaling
Migrations	The Alembic workflow: create → review → apply → roll back
Endpoint reference	The hand-written endpoint list, predating the OpenAPI snapshot
Adaptive sessions	What the Adaptive toggle selects, and what it deliberately ignores
Study recommendations	How focus areas are ranked, and what readiness means
Reranking	Where the cross-encoder is applied and where it deliberately is not
Retrieval thresholds	The measurement behind sourced / adjacent / open
Writing articles	Cross-references, key-point highlights, teaching tips, the three views, linking questions
Linking and cards	When to link and when not to; the small context card
Image overlays	Drawing on a figure
Tied conditions
AI Mode agent	The structured-answer engine behind AI_MODE_ENGINE=agent

These are design notes, not a maintained reference

Several predate changes documented in the topic sections. Where one disagrees with a topic page, the topic page was checked against the code more recently. Endpoint reference in particular is superseded by the generated API reference.