The export file format

This describes the file Export everything produces, so another application can read it without guessing.

It is not a LifeWell format. JSON is JSON, and the archive is an ordinary zip of ordinary CSV files.

The three files

Format	File name	Media type	Complete
JSON	lifewell-export-<YYYY-MM-DD>.json	application/json	Yes
CSV	lifewell-export-<YYYY-MM-DD>.zip	application/zip	Yes
PDF	lifewell-export-<YYYY-MM-DD>.pdf	application/pdf	No — see below

The date is your own local day, not a server day. The name stays ASCII and the same shape in every language, because it is something you will search a downloads folder for.

The JSON document

One object. A small header, then one key per table, each holding an array of rows.

{
  "exportedAt": "2026-09-08T11:04:22.417Z",
  "schema": 1,
  "range": { "from": "2026-01-01", "to": "2026-09-08" },
  "rowCap": 5000,
  "truncated": ["metric_entries"],
  "profile": { "id": "…", "display_name": "…" },
  "conditions": [],
  "medications": [ { "…": "…" } ]
}

Header field	Type	Meaning
exportedAt	ISO 8601 string, UTC	When the file was made
schema	number	The format version. Currently 1
range.from / range.to	ISO date or null	The window you chose. null means open-ended
rowCap	number	The per-table ceiling that was applied — 5000
truncated	array of table names	Tables that hit the ceiling. Empty means nothing was cut
profile	object or null	Your profile row, carried whole rather than as a one-row array

Three things worth knowing before you write a reader:

• A table with no rows is an empty array, not an absent key. "meals": [] is a statement; a missing meals key would be a question.
• truncated is the only signal that a file is partial. Read it before you conclude somebody has no history.
• The JSON is pretty-printed, two spaces. It is a file you are meant to open and search, and a single eight-megabyte line is not something a text editor opens gracefully.

The CSV archive

A zip holding one .csv per table, named <table>.csv. The columns are the table's own columns.

Detail	Value
Line endings	CRLF, per RFC 4180
Encoding	UTF-8 with a byte-order mark
Quoting	RFC 4180 — doubled quotes, fields containing ", ,, CR or LF are quoted
Empty cell	null and undefined both become an empty cell
Nested values	Kept whole as JSON in the cell, rather than flattened across rows

Two deliberate choices that will otherwise look like bugs:

• A table with no rows still gets a file, carrying only its header row. A missing file is indistinguishable from a failed export; a header alone proves the table was considered.
• A cell whose text begins =, +, -, @, a tab or a carriage return is prefixed with an apostrophe. Spreadsheets treat those characters as the start of a formula. Your data is your own, so this is not protection from a stranger — it is protection for whoever you hand the file to, who on this page is often a doctor. Strip the leading apostrophe if you are parsing rather than opening.

The byte-order mark is there because Excel on Windows reads an unmarked UTF-8 file as the local code page, which mangles any name with an accent in exactly the tool this format exists for.

Every table in the file

Sixty-five tables, grouped by the area they belong to.

• Range — Dated rows honour the date window you chose. Standing rows are always included, because a dose with no medicine attached reads as data loss.
• In the PDF — Listed in full prints the rows. Counted, not listed prints the count and the span but not the rows. Not in the PDF means exactly that.

Health

Table	What it holds	Range	In the PDF
conditions	Conditions	Standing	Listed in full
condition_episodes	Episodes	Dated	Listed in full
medications	Medications	Standing	Listed in full
medication_doses	Medication doses	Dated	Counted, not listed
metric_entries	Tracker readings	Dated	Counted, not listed
user_metric_settings	Tracker settings	Standing	Listed in full
metric_definitions	Tracker types	Standing	Listed in full
meals	Meals	Dated	Counted, not listed
saved_meals	Saved meals	Standing	Not in the PDF
saved_meal_items	Saved meal items	Standing	Not in the PDF
eating_windows	Fasting window	Standing	Listed in full
shopping_lists	Shopping lists	Standing	Not in the PDF
shopping_list_items	Shopping list items	Standing	Not in the PDF
movement_sessions	Workouts	Dated	Counted, not listed
wellness_breaks	Wellness breaks	Dated	Not in the PDF
wellness_nudges	Wellness nudges	Standing	Not in the PDF

Mind

Table	What it holds	Range	In the PDF
journal_entries	Journal entries	Dated	Not in the PDF
mind_sessions	Guided sessions	Dated	Counted, not listed

Life stages

Table	What it holds	Range	In the PDF
cycle_days	Cycle days	Dated	Counted, not listed
babies	Children	Standing	Listed in full
baby_growth	Growth measurements	Standing	Listed in full
baby_logs	Baby logs	Dated	Counted, not listed
baby_milestones	Milestones	Standing	Listed in full

People

Table	What it holds	Range	In the PDF
people	People	Standing	Listed in full
person_edges	Relationships	Standing	Not in the PDF
connections	Connections	Standing	Listed in full
connection_prefs	Connection settings	Standing	Not in the PDF
share_links	Shares	Standing	Listed in full
share_submissions	Sent through a share link	Standing	Not in the PDF
community_members	Groups you joined	Standing	Not in the PDF
posts	Community posts	Standing	Listed in full
post_replies	Community replies	Standing	Not in the PDF
blocks	People you blocked	Standing	Not in the PDF
chats	Conversations	Standing	Not in the PDF
chat_members	Who is in each conversation	Standing	Not in the PDF
messages	Messages you sent	Standing	Not in the PDF
message_hides	Messages you hid	Standing	Not in the PDF
professionals	Your professional listing	Standing	Listed in full
professional_claims	What you told us about yourself	Standing	Not in the PDF
professional_slots	Times you offered	Standing	Not in the PDF
professional_verifications	Register checks	Standing	Not in the PDF
blood_donor_profiles	Your blood donor listing	Standing	Listed in full
blood_requests	Blood requests you sent	Standing	Listed in full

Memories

Table	What it holds	Range	In the PDF
memories	Memories	Standing	Not in the PDF
scrapbook_items	Scrapbook items	Standing	Not in the PDF
media_files	Media items	Standing	Counted, not listed
media_albums	Albums	Standing	Listed in full
media_people	Who is in a photo	Standing	Not in the PDF
tags	Tags	Standing	Listed in full
taggables	Tagged items	Standing	Not in the PDF

Everyday

Table	What it holds	Range	In the PDF
notes	Notes	Standing	Not in the PDF
note_folders	Note folders	Standing	Not in the PDF
location_points	Saved places	Standing	Counted, not listed
location_settings	Location settings	Standing	Listed in full
events	Calendar	Dated	Listed in full
event_checklist_items	Appointment notes	Standing	Not in the PDF
reminder_schedules	Reminders	Standing	Listed in full

Your account

Table	What it holds	Range	In the PDF
profiles	Your profile	Standing	Listed in full
notifications	Messages sent to you	Dated	Not in the PDF
export_runs	Earlier exports	Standing	Not in the PDF
export_schedules	Scheduled exports	Standing	Not in the PDF
plan_grant_requests	Plan requests	Standing	Not in the PDF
referrals	People you invited	Standing	Not in the PDF
referral_rewards	What your invitations earned	Standing	Not in the PDF
account_audit	Changes to your account	Dated	Not in the PDF

What each format leaves out

JSON and CSV are complete. Every one of the sixty-five tables above is in both.

The PDF is not, and that is the point of it. A PDF is a document people print, hand to a receptionist and leave on a desk. So it carries what a clinician actually reads and nothing else:

• Listed in full — the standing records: conditions, medications, tracker types and targets, children, growth measurements, milestones, people, connections, shares, calendar, reminders, your profile.
• Counted, not listed — the high-volume dated ones: tracker readings, medication doses, meals, workouts, cycle days, baby logs, guided sessions, media items, saved places. Forty rows out of four thousand is a sample, and a sample in a document labelled your record misleads the one reader it exists for. A listed table prints at most 60 rows.
• Not in the PDF at all — the journal, memories, notes, scrapbook items, messages, community replies, and everything else somebody wrote for themselves.

The journal is never in the PDF, and that one is not a preference

Everything else on this page is a choice about what is useful. The journal is the one surface in LifeWell that must not leave on a sheet of paper somebody sets down on a desk. It is in the JSON and the CSV, which are files you open yourself.

What no format ever contains

Seven things are absent from every export, including the complete ones.

Absent	Why
Media files themselves	The images live in your own Google Drive. The export carries the reference, the caption and the date — the bytes were never ours to hand over
Anything shared with you	Somebody else's record, lent to you. It leaves when they revoke it, which a copy in your file would not
Deleted rows	Deletion is deletion. There is no recycle bin behind the export
Operational logs	Rate-limit counters, dismissed notices and similar. Interface state, not your record
Your password	There is not one. Sign-in is Google, and we never see or store a password
Your AI provider key	Held encrypted, and it is a live credential. Writing it into a plain file you are about to email yourself would be a leak wearing the costume of completeness
AI usage metering	The meter a limit is counted from, not something you wrote

Five further tables are held back for the same class of reason, each of them a credential or somebody else's record:

Held back	Why
Reports you made about a post	Nobody can read that table back, including you. A moderation queue that leaks through a download is the same leak with an extra step
A blood donor's released contact details	Given once, to one person, and withdrawable. A row in a file is neither
Your Google Drive connection	It holds a live refresh token
A Play purchase record	It holds a purchase token Google will honour. Your entitlement is already in the file, on your profile
Dismissed announcements	Which notices you closed. Interface state

And four tables are shared reference data rather than yours: the plan list, the community catalogue, the food database, and one internal table used by a keep-alive check.

Reading it back

There is no import. Nothing in LifeWell reads one of these files back in.

The format is documented so that something else can — a script, a spreadsheet, another application, or you in five years with a text editor. Two properties make that practical, and both are deliberate:

1. Every identifier in the file is the database's own column name, so a column means the same thing in the JSON, the CSV and this page.
2. The header names its own limits. A partial file says it is partial in its first ten lines.

When this page changes

Every wave that adds a table adds its row above, and a test asserts the list here matches the schema in both directions — a table in one and not the other fails the build rather than shipping a silent gap.

If you are writing a reader, key off schema. It is 1 today, and it will increase before anything in the shape above changes meaning.