> ## Documentation Index
> Fetch the complete documentation index at: https://foreverbetter.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Data model

> The five primitives every request flows through: sources, observations, interpretations, analyses, and dashboard specs.

Every modality shares one canonical shape, so an app or agent learns the model
once and reuses it across biomarkers, wearables, genetics, and behavioral data.

## The primitives

| Primitive          | What it is                                                                                                                       |
| ------------------ | -------------------------------------------------------------------------------------------------------------------------------- |
| **Source**         | A raw upload or connected pull (`RawSourceReference`). Preserves filename, provider, byte length, and storage mode.              |
| **Observation**    | A normalized reading extracted from a source (`NormalizedObservation`): a lab value, a wearable metric, a supplement, a symptom. |
| **Interpretation** | A derived finding (`DerivedInterpretation`): a scored status, a derived metric, or a genetic summary, always with provenance.    |
| **Analysis**       | A run that combines sources into interpretations plus a healthspan score (`AnalysisResult`).                                     |
| **Dashboard spec** | The frontend-ready JSON: cards, summaries, scores, provenance.                                                                   |

## Provenance is always attached

Every interpretation records `source_ids`, `source_categories`, a `source_type`
(`direct`, `derived`, `combined`, `queued`, `setup_required`, `failed`), the
`engine` that produced it, and a timestamp. Derived metrics also record their
`inputs`. This lets a customer trace any number on a dashboard back to the exact
upload it came from.

## Modalities

`biomarkers`, `wearables`, `genetics`, and `behavioral`. One modality is useful
on its own; the unified health context reports which modalities are present,
what is missing, and the impact of each gap.

## Discover without bookkeeping

* `GET /capabilities`: supported modalities, providers, and setup state.
* `GET /sources` and `GET /analyses`: everything a user has uploaded and computed.
* `POST /users/{user_id}/health-context`: a bounded, provenance-rich summary
  across all modalities.