# Ask about your data

Keep the location points you explicitly record. Export coordinates and their original accuracy, altitude, and timing fields. Points recorded by myself.md through Expo Location.

Copy a question into your agent to get started. This catalog describes capabilities, not a connected person's available records. All answer examples are fictional.

## Where did I record locations last week?

Review the location points you previously chose to record with myself.md.

Fictional answer example: Your recorded points form two clusters of coordinates during the week. There are no points for Tuesday, and I cannot confirm where you were that day.

Data needed: Points recorded by myself.md during the selected dates.

Availability: iOS and Android, after background recording is enabled with the required location permission.

- myself.md cannot retrieve a historical location archive from the OS; recording must already have been running.
- Recorded points have accuracy and sampling gaps. They do not establish a complete route or confirmed place visits.

Prompt to send to your agent:

Use myself.md: Where did I record locations last week? Summarize recorded points and their accuracy, without inferring a complete route or confirmed visits.

Use approved records, confirm dates and timezone, and explain missing data. If needed, help me connect https://myself.md/mcp using https://myself.md/docs and wait for my approvals.

[Explore location data types and example JSON](/datasets/location#technical-explorer).

## Connect your data when you’re ready

1. Connect a compatible AI assistant at https://myself.md/mcp and complete sign-in.
2. Pair an installed myself.md phone app with the same account. Public store downloads are coming soon.
3. Select the required data in a phone profile, allow the relevant OS permissions and separately approve its Agent permissions.

Viewing this page or copying a prompt does not grant access. Your assistant must check available records and date coverage before answering. Public App Store and Google Play downloads are coming soon.

[Connect an AI assistant](/docs). Agents can discover these same structured workflows, required types, prerequisites and query steps through `get_public_dataset_catalog` or [the public catalog](/v1/dataset-catalog). Follow `nextCursor` until `hasMore` is false.

1. `get_phone_data_catalog`: Discover connected phones, approved profiles and actual supported types. Catalog support does not prove that records exist.
2. `query_phone_data`: Use the discovered device/profile IDs and approved native types. Resolve the question to explicit date bounds and a timezone; request only the needed data.
3. `get_phone_request`: Poll queued or running requests. Use records only after complete; report failed, expired or missing requests. Follow every result cursor with a new bounded query.

## Technical explorer

Location exports in myself.md. Select native types in the interactive explorer to update a fictional JSON export. Agent reads require supported phone sources, operating-system permission, selected profile types and separate agent approval.

## Location: sources and supported collections

Keep the location points you explicitly record. Export coordinates and their original accuracy, altitude, and timing fields.

Sources: Points recorded by myself.md through Expo Location.

- iOS and Android · Recorded data: 1 selectable types.

## Location: platform limitations

- On iOS, Allow Once or denied Always access can require a trip to system settings.
- The OS does not expose a historical location archive to this adapter. Only points recorded by myself.md are available.
- Background recording requests highest accuracy, a 5 metre distance interval, and a 10 second time interval. These are requested settings, not a guaranteed sampling rate.
- If background access is denied, tracking does not start.
- Stopping tracking stops future recording; previously saved points remain. Local queries include records whose start is inside the UTC interval.

## Location: fictional native record example

This illustrates native fields for points from expo-location, not a complete export envelope or a real person's records. The interactive builder previews selected types in the released export schema.

```json
{
  "timestamp": 1791446400000,
  "coords": {
    "latitude": 38.7223,
    "longitude": -9.1393,
    "accuracy": 8,
    "altitude": 24,
    "altitudeAccuracy": 10,
    "heading": null,
    "speed": null
  }
}
```

[Explore location types and build a fictional export](/datasets/location).

## Connect an AI assistant

Add https://myself.md/mcp as a remote MCP server in a compatible client and complete OAuth sign-in. Pair your phone with the same account, select a profile and approve its relevant individual Agent permissions. The assistant discovers available types before requesting a bounded query and waits for completion. A stored cloud export is a separate read path; signing in or viewing this catalog grants no private data.

Read [getting started](/docs), the [API reference](/docs/reference), [pricing](/pricing) and [privacy](/privacy).

For source-specific instructions, read [Apple Health with Claude and scoped MCP access](/docs/apple-health-claude-mcp) or [Android Health Connect exports and MCP queries](/docs/android-health-connect-mcp).

## Explore other datasets

- [Health](/datasets/health): HealthKit on iOS, Health Connect on Android, and optional WHOOP on either phone platform.
- [Screen time](/datasets/screen-time): The PhoneUsage native adapter on iOS and Android.
- [Location](/datasets/location): Points recorded by myself.md through Expo Location.