Positive Experience Format

P.E.A. holds positive experiences from any source: check-in apps like Swarm, journals like Day One, or a table from a spreadsheet. Your data stays on your devices and in your personal iCloud account. This page describes the file P.E.A. exports and imports, the Positive Experience Format (P.E.F.).

What this page does

It documents the data in a P.E.A. backup file (.json), so you can:

• Read your experiences in any text editor

• Move your data between iCloud accounts, or merge two archives

• Bring data in from another app, by hand, by script, or with an AI model

• Move your data into another app

The format is versioned and public. This page describes format v4 (exportSchemaVersion 4) and was last updated on November 1, 2026. Every version since the first is in the Format Version History.

Format structure

Every backup or export is one JSON file, usually named like PEA-Export-2026-05-12.json. It is plain text, a JSON object that looks like this:

{
  "exportSchemaVersion": 4,
  "exportDate": "2026-05-23T17:30:42Z",
  "appVersion": "1.0",
  "entryCount": 247,
  "placeCount": 12,
  "mottoCount": 1,
  "peopleCount": 18,
  "conceptCount": 34,
  "readme": "P.E.A. — Positive Experience Archive\n…",
  "entries":  [ … ],
  "places":   [ … ],
  "mottos":   [ … ],
  "people":   [ … ],
  "concepts": [ … ]
}

• exportSchemaVersion sets the rules P.E.A. reads the file by. A file with a version it doesn't know is refused, not guessed at.

• entries, places, mottos, people and concepts are the top-level collections. (concepts holds the Topics: the app called them Concepts when the format was published, and a key in a public format never changes, so older backups keep working.) Every object has a uuid, so a later import recognizes it instead of making a duplicate. A file with remembered topic suggestions also has topicAssociations and topicAssociationCount (v4). people, concepts, topicAssociations and their counts are left out of a file that has none of them.

• An entry with a saved place includes that place's fields (placeName, placeDetail, placeSource, and so on), so it makes sense on its own. People and topics go the other way: entries point at them by uuid (peopleUUIDs, conceptUUIDs), and the top-level people and concepts lists hold the records.

• Optional fields are left out when there is no data. No null, no empty strings. That keeps the file easy to read and edit.

Positive Experience Archive Schema

Every experience is one object in the entries array. Required fields are marked R.

Identity and content

Time

Your physical location at capture

P.E.A. never edits a captured value, and it fills in the empty ones after import: an entry with coordinates gets its city, neighbourhood, county, postal code, region, country and time zone, and the day's weather and moon where the record has them, so bring the coordinates, leave out what you don't know, and never guess.

A place you want to reference (semantic place)

A saved place you assigned to the experience, or that P.E.A. assigned because the experience happened within the place’s radius (100 m unless you changed it); these optional fields mirror a row in the top-level places collection and never overwrite the location where the experience was captured.

Weather and sky

Leave these out when your source doesn't have them: after import, P.E.A. recovers the hour's condition and temperature and the day's moon phase for entries with coordinates, where the record covers them, unless weather capture is turned off in the app.

Activity, motion, environment

Music

Capture source device

Source provenance and identity (v2)

Which app the entry came from and its stable identifier there. On Add data to P.E.A., a row with the same (source, sourceID) as one already in the archive, or as an earlier row in the same file, is skipped; so is a row whose createdAt and text both match one, and the preview says so before you confirm. Once set, these values never change, through every export and re-import.

Intensity state (v2)

People links (v2)

Topic links (v3)

Soft-deletion (Neutralization)

A neutralized experience leaves the list, stays in the Index under the Neutralized filter for 30 days, and is then deleted, or earlier when you delete it yourself.

Place Schema

Places you save and reuse (“Home”, “Jan’s Haus”, “Papa’s”); every place is one object in the top-level places array, your list of saved places, while an entry’s own place fields say which place is assigned to that experience. Required fields are marked R.

websiteURL through country are the place’s listing (v3.6): what Apple Maps said about the venue when you linked it, kept as recorded and never refreshed, so the place card can show an address, a website and a phone number without going online. detail stays the short “City, Country” line. A place P.E.A. never looked up has no listing, and a listing that only had an address has no website or phone.

On Add data to P.E.A., a place is a duplicate when its uuid matches, or when its name and both coordinates match to five decimals (about a meter), because the same place has different UUIDs on different devices. Two places with the same name at different coordinates stay two places, and both keep their name. Names come through exactly as they are in the file, in both modes.

Place names are not unique. One archive can hold several places called "Home". Code against this format keys places by uuid, or by identityKey where it is there, never by name.

Place identity (v3.3)

A stable key for a place that doesn’t depend on its name; every field here is optional, added at v3.3 without a version change, and a file without any of them is still valid.

Identity keys are opaque: never read a coordinate or an Apple ID out of one, never make one up, and never copy one onto a different place (two places with one key are one place). Leave them out and P.E.A. works out the identity on import.

Motto Schema

Three words you entered or picked that show before archiving, on every device; every motto is one object in the top-level mottos array.

Mottos are off until you turn them on in the app; importing an archive with mottos turns them on. On Add data to P.E.A., a motto is a duplicate when its uuid matches or when all three lines match exactly: mottos are typed by hand and have no other identity, and the same motto has different UUIDs on different devices. inactive is a state, not part of the match, and a merge never flips it on a motto you already have.

People Schema (v2)

People you associate with an experience in the editor's People field, each one object in the top-level people array, pointed at from entries through peopleUUIDs; required fields are marked R.

On import, P.E.A. creates every person first and then connects the entries, so every UUID finds its row. On Add data to P.E.A., a person is a duplicate when the uuid matches, or when source and sourceID both match; people made in P.E.A. have no sourceID, so for them only the uuid counts. Archiving on the iPhone, the Mac, the watch, the widget or through Siri never creates a person, but when the text names someone already in People, P.E.A. associates them on arrival; People have their own page under Manage, a filter in the Index and on the Map, and a grouping in the Map’s list. The Swarm converter creates no people, because Foursquare’s export leaves out who you were with; the only trace is an @mention in the shout text.

P.E.A./Topic Schema (v3)

A topic is a word or two that says what an experience was about (“soccer”, “made”, “first time”). You create and associate them in the entry editor’s Topics field: tap a suggestion, a default seed, or type a new one. On screen they are Topics; in the file they are concepts, because the app called them Concepts until September 2026 and a key in a public format doesn’t change. A converter writes concepts, conceptUUIDs and conceptCount, each topic one object in the top-level concepts array, in the same shape as a person. People answer who, places answer where, topics answer what.

On import, P.E.A. creates every topic first and then connects the entries. On Add data to P.E.A., a topic is a duplicate when the uuid matches, or when source and sourceID both match; topics made in P.E.A. have no sourceID, so two archives that each made a "made" topic end up with two.

Topic Suggestions Schema (v4)

What P.E.A. remembers about suggesting topics: the topics you keep using at a place, the words you picked for a kind of place or an activity, the suggestions you turned off, and the rule always for a topic a place adds by itself. Each link is one object in the top-level topicAssociations array, counted by topicAssociationCount. The topics on your experiences stay on the entries as conceptUUIDs, and an archive without this array imports as before, only without the remembered suggestions. Required fields are marked R.

Every object has a topicUUID, a starterID, or both. On Add data to P.E.A., an object matches one already in the archive when targetKind, targetKey and the topic or word match, and the newer updatedAt wins, so a suggestion you turned off on your phone doesn't come back from an older file. A topicUUID that is in neither the file nor the archive is kept and does nothing until that topic shows up. A converter with no preferences to write leaves the array out.

P.E.F./One Experience example in detail

An experience archived in P.E.A. itself (source: "pea"), with a real archiveNumber and the intensity the person chose, as it looks in an export:

{
  "exportSchemaVersion": 4,
  "exportDate": "2026-05-12T17:30:42Z",
  "appVersion": "1.0",
  "entryCount": 1,
  "placeCount": 1,
  "mottoCount": 0,
  "peopleCount": 0,
  "conceptCount": 0,
  "readme": "P.E.A. — Positive Experience Archive\n=====================================\n…",
  "entries": [
    {
      "uuid": "1A2B3C4D-5E6F-7081-9202-A1B2C3D4E5F6",
      "archiveNumber": 412,
      "text": "Falco on Walmart radio.",
      "intensity": 2,
      "intensityLabel": "Light",
      "isIntensityUnset": false,
      "createdAt": "2026-01-15T23:32:00Z",
      "updatedAt": "2026-01-15T23:32:00Z",
      "timeOfDay": "Evening",
      "dayOfWeek": "Thursday",
      "season": "Winter",
      "timeZoneID": "America/New_York",
      "latitude": 35.2271,
      "longitude": -80.8431,
      "locationName": "Walmart Supercenter",
      "locationLocality": "Charlotte",
      "locationAdministrativeArea": "NC",
      "locationSubLocality": "Wesley Heights",
      "locationSubAdministrativeArea": "Mecklenburg County",
      "locationPostalCode": "28208",
      "locationCountry": "United States",
      "placeName": "the Walmart by the airport",
      "placeDetail": "Charlotte, NC",
      "placeSource": "custom",
      "placeLatitude": 35.2270,
      "placeLongitude": -80.8430,
      "placeMapItemIdentifier": "I3F4A2C8B9D0E1F2A3B4C5D6E7F8A9B0",
      "placeCategory": "Store",
      "isUserConfirmedPlace": true,
      "weatherCondition": "Clear",
      "temperatureCelsius": 7,
      "moonPhase": "Waxing Crescent",
      "activityType": "walking",
      "speedMPS": 0.6,
      "altitudeMeters": 226,
      "stepCountToday": 6428,
      "noiseDecibelLevel": 68,
      "nearbyPOI": "Walmart Supercenter, Bojangles, ExxonMobil",
      "musicTrack": "REAL ONES NEVER DIE",
      "musicArtist": "Kid Cudi",
      "musicStoreID": "1886092613",
      "deviceName": "iPhone 15 Pro",
      "deviceType": "iphone",
      "source": "pea"
    }
  ],
  "places": [
    {
      "uuid": "9F8E7D6C-5B4A-3928-1706-D5C4B3A29180",
      "name": "Airport Walmart",
      "detail": "Charlotte, NC",
      "latitude": 35.2270,
      "longitude": -80.8430,
      "mapItemIdentifier": "I3F4A2C8B9D0E1F2A3B4C5D6E7F8A9B0",
      "createdAt": "2025-12-30T14:10:00Z",
      "websiteURL": "https://www.example-store.com/charlotte",
      "phoneNumber": "(704) 555-0148",
      "street": "3850 E Independence Blvd",
      "locality": "Charlotte",
      "administrativeArea": "NC",
      "postalCode": "28205",
      "country": "United States"
    }
  ],
  "mottos": [],
  "people": [],
  "concepts": []
}

What happened at this Walmart: the person was listening to Kid Cudi when the store's radio played Falco, and that was worth a note. P.E.A. kept both, the song on the headphones and the note about the other one. The coordinates, the weather, the moon, the steps and the song come from the iPhone; placeName is a place the person saved. Nothing captured is edited afterwards, and what wasn't available is left out. When you create or edit archive files, follow the same rule. Intensity 2, Light: Falco at Walmart is not earth shattering, but it is weird and fun, and that is what P.E.A. is built to save.

P.E.F./Import + export

P.E.A. imports only its own format (JSON), and any tool can produce it, including a chat with an AI model. Convert your data into one PEA-Export-….json file and add it through Manage → Import & Export → Add data to P.E.A. (on the Mac, File → Add Data to P.E.A.…, then Choose File…). That is a merge: everything you already have stays as it is.

Imported entries arrive Unset. P.E.A. never guesses how strong a moment was. Every converted row comes in as a light-gray Unset dot with no P.E. # number. You rate it by dragging the dot in the app, and that is when it gets its permanent number. A row you never rate stays in your archive, numberless, for as long as you keep it; nothing deletes it for you. The prompt below sets every row to Unset and leaves the rating to you.

Import prompt

Paste this prompt with your entries into a chat model, or into the Use Model action in Shortcuts. The rules are built in, so it also works with models that can't open links: Apple Intelligence on-device, Ollama, LM Studio. A dozen entries at a time on-device, more through Private Cloud Compute or a chat model. Make sure you trust the model you use and its data policy; a free service likely keeps your data.

Turn the entries below into one P.E.A. archive file. Answer with the
JSON only. The full schema is at https://archive.green/format, but
everything you need is here.

The file is one object with exactly these keys:
"exportSchemaVersion": 3, "exportDate" (now, ISO 8601, like
"2026-10-07T09:00:00Z"), "appVersion": your converter's name and
version in one string, like "notes-to-pea/1.0" (a bare "1.0" reads
as a P.E.A. backup, and a name tells P.E.A. the entries were
converted, so it dates their arrival to the import), "readme": "Converted
for P.E.A.", "entryCount", "placeCount", "mottoCount", "peopleCount",
"conceptCount", "entries", "places", "mottos", "people", "concepts".

One object in "entries" per entry:
- "uuid": a new random UUID (8-4-4-4-12 hex), different for every entry,
  unless you are reusing the ids of a P.E.A. archive
- "archiveNumber": 0 on every row. P.E.A. gives a row its permanent
  P.E. #N when I first rate it in the app. Never pre-number a row.
- "text": the entry's own words, or "" if there are none
- "intensity": 1, "intensityLabel": "Minimum", "isIntensityUnset": true
  on every row. Never guess a strength from the text; I rate each
  moment myself.
- "createdAt" and "updatedAt": the entry's date and time, ISO 8601 with
  the time zone (like "2025-04-29T20:05:00-07:00"), the same value in
  both. If only the day is known, use 12:00.
- "timeOfDay" from the local clock: "Morning" before 12:00, "Afternoon"
  from 12:00, "Evening" from 17:00, "Night" from 21:00
- "dayOfWeek": "Monday" to "Sunday"
- "season" from the month and the hemisphere. North of the equator:
  "Spring" for March to May, "Summer" for June to August, "Autumn" for
  September to November, "Winter" for December to February. South of
  it: "Autumn" for March to May, "Winter" for June to August, "Spring"
  for September to November, "Summer" for December to February.
  Without coordinates, use the northern table.
- "source": a short lowercase word for where the data came from
  ("dayone", "notes", "csv"). Never "pea".
- "sourceID": the entry's id in the source, when it has one
- "timeZoneID": the IANA zone ("Asia/Tokyo"), only when the source
  records it. Never guess a zone from an offset.
- Only when the entry says so: "latitude" and "longitude" as numbers,
  "locationName" (the place's name), "locationLocality" (the town),
  "locationAdministrativeArea" (the state or region), "locationCountry".

People and topics: for every person named, one object in "people" with
"uuid", "name", "createdAt", "updatedAt" (same moment as "exportDate")
and the same "source". Put that person's uuid in the entry's
"peopleUUIDs". Topics go the same way into "concepts" and
"conceptUUIDs". Use one uuid per name, however often it comes up.

Saved places and mottos, only when the source has them: a place is one
object in "places" with "uuid", "name", "createdAt", and "latitude" and
"longitude" when known; write its address, website or phone only when
the source records them, never looked up, because P.E.A. fills in a
place's listing itself after import. A motto is one object in "mottos"
with "uuid", "line1", "line2", "line3", "createdAt". When the source has
none, "places" and "mottos" are empty arrays.

Leave out every field you don't have. Never write null. Don't guess a
place, a time zone, the weather, the moon or a rating: P.E.A. fills in
the city, region, country, time zone, weather and moon from the
coordinates after import. Every count ("entryCount", "placeCount",
"mottoCount", "peopleCount", "conceptCount") is the length of its
array.

Entries:
[paste your notes, your journal export or your spreadsheet rows]

Save the answer as a .json file and add it through Manage → Import & Export → Add data to P.E.A. Every entry arrives Unset. Rate it, and it gets its number.

Keep your data private

Your data is yours, and converting it should not change that.

• Apple Intelligence on iOS 26 / macOS 26 (Foundation Models, Writing Tools, the Shortcuts "Use Model" action) runs on-device when it can and uses Private Cloud Compute for longer inputs, where nothing is kept and nobody at Apple can look.

• A local model through Ollama, LM Studio or an MLX runner: a 7B–8B model (Llama 3.1, Qwen 2.5, Gemma 3, Mistral) on a Mac with 16 GB converts a year of entries in a few minutes. No account, nothing goes over the network.

• Hosted APIs (OpenAI, Anthropic) have no-training and no-retention modes. Check the settings and decide whether you trust the vendor.

• Avoid free consumer chat apps, where the data settings are easy to get wrong. Entries hold exact locations and personal moments.

Sources we expect people to bring

• Swarm / Foursquare check-ins: the converter script and the field mapping are at archive.green/swarm-to-pea. Importing doesn't mean leaving Swarm: P.E.A. is not a replacement for it, but if your good times are in your check-ins, this is a place to keep them for good.

• A spreadsheet: the CSV template and its prompt are at archive.green/import-template.

• Day One JSON: detailed times and locations; a model converts it cleanly with the prompt above.

• Apple Notes or Markdown journals: the text and the dates are enough to start. Everything else is optional.

Restoring Safely

P.E.A. has two ways to bring data in:

• Add data to P.E.A. is a merge. New rows are added. Rows that already exist (same uuid, same (source, sourceID) for imported data, or the same createdAt and text), or that an earlier row in the same file already covers, are skipped, so one moment under two ids lands once, and the preview tells you before you confirm. This is for converted imports, Swarm, Day One, a spreadsheet, and for adding an existing P.E.A. backup to what you have.

• Restore P.E.A. from backup replaces your current data. It is for your own P.E.A. exports (moving to a new iCloud account, recovering after a wipe). It is destructive. It only accepts archives that P.E.A. exported.

The format version is independent of the P.E.A. app version. It is a plain integer that only changes when the archive's shape changes. P.E.A. always reads older versions; newer files are refused with a clear message until you update the app. Every version is in the Format Version History.