Daily records

Dieser Inhalt ist noch nicht in deiner Sprache verfügbar.

A Health.md daily record combines a readable Apple Health summary projection with optional typed provider data and source-level Apple Health records.

healthmd.health_data v8
├── identity, date, timezone, and units
├── optional Apple summary sections
├── optional providers.whoop (healthmd.provider.whoop_daily v1)
├── raw capture status
├── optional healthkit_record_archive
└── optional exporter diagnostics

Use summaries for dashboards and journaling. Use the canonical archive for source identity, exact measurements, provenance, and relationships.

Top-level fields

Field	Type	Presence	Meaning
`schema`	string	Always	`healthmd.health_data`.
`schema_version`	integer	Always	Current Apple daily contract version, `8`.
`date`	`YYYY-MM-DD` string	Normal successful document	Calendar day represented by the summary.
`type`	string	Always	Normally `health-data`; serialization-error fallback documents use a distinct value.
`time_context`	object	Always	Captured calendar timezone and machine timestamp timezone.
`unit_system`	string	JSON	Structured v8 JSON currently reports `metric`; display preferences affect prose, not canonical structured values.
`units`	object	JSON/frontmatter	Exported summary key to canonical unit. Empty/list-like fields are omitted.
`raw_capture_status`	enum string	Always	`complete`, `partial`, `not_requested`, or `legacy_unavailable`.
Summary sections	object/array	Conditional	Emitted only when the selected section has retained data.
`providers`	object	Conditional	Typed provider namespace. Apple v8 currently permits `providers.whoop` v1.
`healthkit_record_archive`	object	Conditional	Present when source capture was requested and available.
`diagnostics`	object	Conditional	Export/fetch partial failures outside the archive query manifest.

The exact generated top-level paths and observed types are listed in generated/core/daily-json-fields.md.

Time context

{
  "time_context": {
    "calendar_timezone": "UTC",
    "timestamp_timezone": "UTC"
  }
}

calendar_timezone is the IANA timezone used to assign the daily date and render short clock values.
timestamp_timezone is always UTC for complete machine timestamps.
Canonical archive timestamps have a fixed nine-digit fractional component.
Summary compatibility windows can differ from canonical ownership. Sleep summaries retain Health.md’s established night-oriented attribution, while raw records belong to the day containing their source start date.

Summary sections

Daily JSON may emit these sections. Fields are omitted when no retained value exists.

Section	Examples of data represented	Notes
`sleep`	Duration, stage durations, bedtime/wake, stage samples	Readable projection; raw sleep categories remain in the archive.
`activity`	Steps, energies, exercise, stand time/hours, distances, VO2 Max	VO2 Max includes source provenance and carry-forward state when available.
`heart`	Resting/walking/average/min/max rate, HRV, samples	Summary values can be aggregates over many source samples.
`vitals`	Respiratory rate, oxygen, temperature, blood pressure, glucose, samples	Blood-pressure truth is the canonical correlation graph, not inferred sessions.
`body`	Weight, height, BMI, body fat, lean mass, waist	Canonical structured units are stable.
`nutrition`	Energy, macronutrients, water, caffeine, fats	Nutrient totals are projections over selected source records.
`mindfulness`	Mindful activity and State of Mind views	Source UUIDs and exact labels/associations are retained in canonical records.
`mobility`	Walking, stairs, running mechanics, six-minute walk	Values use canonical units independent of display preference.
`hearing`	Headphone and environmental sound levels	Additional hearing events may be archive-only.
`reproductiveHealth`	Selected reproductive summary values	Many newer reproductive/pregnancy definitions are archive-only.
`cyclingPerformance`	Speed, power, cadence, FTP	Workout-associated samples also appear in workout graphs.
`vitamins`	Every selected vitamin total	Microgram keys retain the canonical `µg` unit introduced in the v7 dictionary.
`minerals`	Every selected mineral total	Microgram and milligram nutrients remain distinct.
`symptoms`	Symptom occurrence counts	Source categories and raw values live in the archive.
`medications`	Inventory summaries and dose-event projections	Canonical dose events and inventory identities are retained separately.
`other`	UV, daylight, falls, alcohol, insulin, hygiene, water/depth	Exact availability depends on OS and authorization.
`workouts`	Identity, type, duration, distance, rates, laps, splits, route, series	The canonical workout graph is authoritative for complete public source data.

WHOOP data remains provider-namespaced and never populates these Apple summary sections. Its exact nested paths are generated from generated/core/provider-day.json; repeated records remain structured, while unambiguous daily scalars use whoop_* keys in Bases/frontmatter and CSV.

The exhaustive metric-to-summary-key table is generated at generated/core/metric-catalog.md. It includes every selected metric definition, category, HealthKit identifier, source aggregation, canonical summary keys, units, and archive-only status. generated/core/metric-examples.md provides one synthetic, source-derived reference entry for every metric ID.

Omission and null rules

Health.md favors omission over fabricated values:

A missing metric is not exported as zero.
An empty optional summary section is omitted.
A missing source/device field is omitted from the canonical object.
A successful empty source query is represented in the query manifest with record_count: 0; it is not converted into a fake sample.
UUID-free public values do not receive fabricated HealthKit UUIDs.
Unavailable binary bytes do not receive fabricated data or checksums.

Downstream databases should distinguish missing, zero, false, and empty list.

Summary keys versus JSON fields

Health.md exposes two structured summary styles:

Nested JSON fields such as activity.activeCalories.
Flat frontmatter/Bases keys such as active_calories.

When Write Data Dictionary is enabled, the data dictionary maps flat keys to metric IDs, units, HealthKit identifiers, and aggregation rules. A flat key can be renamed by the user; its canonicalKey remains stable in the dictionary. Consumers must tolerate the sidecar being intentionally absent.

JSON field names and flat frontmatter names are not interchangeable. Use the generated references for each format rather than guessing a conversion.

Units

Structured output uses stable units. Examples include:

distances in explicit meter/kilometer/mile fields;
body mass in kilograms;
temperatures in degrees Celsius;
percentages either as documented ratios or explicit percent fields;
Stand Time in minutes and Stand Hours as a distinct hour count;
microgram summary nutrients as µg, matching the production v8 dictionary and CSV Unit cells;
exact reviewed HealthKit query units inside source quantity payloads, including mcg for microgram source records.

Markdown prose can use the selected Metric or Imperial display preference. Never infer a structured unit from rendered prose.

Capture-state examples

Lossless complete

raw_capture_status: complete with an archive means all planned supported branches completed. The archive can legitimately contain zero records.

Lossless partial

raw_capture_status: partial means at least one requested branch failed, was cancelled, was skipped, or was unsupported. Successful sibling records remain usable. Inspect the manifest before accepting the day as complete.

Summary-only

raw_capture_status: not_requested means Lossless Health Records was explicitly off for this export. No canonical archive is present.

Legacy peer or persisted record

raw_capture_status: legacy_unavailable means the source predates canonical archive support or arrived from a peer that could not provide it.

Complete synthetic examples

These files are generated by production exporters. They are examples, not templates to mutate into a different schema.