This documentation is available as Markdown for AI agents and LLMs. See the full Markdown index or append .md to any documentation URL.

Query EAS Observe metrics, events, and sessions from the terminal with the eas observe commands.

Everything the EAS Observe dashboard shows is also available from the terminal. Use the eas observe commands to compare releases, investigate slow sessions, and pipe results into scripts.

Prerequisites

EAS CLI help

Run any command with --help to see the flags supported by your installed EAS CLI version.

Some data is available only on certain plans. When your account's plan does not include what a command asks for, the command fails with an upgrade message that links to your billing page. Session timelines are checked before the interactive picker runs, so a blocked plan is reported immediately. See Pricing for what each plan includes.

Commands

Every command accepts these flags:

• --platform android or --platform ios: filter by platform. Both are included by default. Not available on observe:session, which already targets one session.
• --days <number>: show data from the last N days.
• --start <ISO date> and --end <ISO date>: set an explicit time range. Mutually exclusive with --days.
• --project-id <id>: query a project without running inside its directory.
• --json: machine-readable output. Implies --non-interactive.
• --non-interactive: fail instead of prompting.

When no time range is given, commands return the last 60 days.

Metric names

Startup metrics are collected automatically once the app is instrumented. See the Metrics reference for what each one measures.

Navigation metrics are per route. They require SDK 56 or later and one of the navigation integrations, either Expo Router or React Navigation.

observe:metrics and observe:metrics-summary accept all nine names. observe:routes accepts the three navigation names.

eas observe:metrics-summary

Shows aggregated statistics grouped by app version, with a separate table per platform. Use it to compare startup performance across releases.

Command flags:

• --metric <name>: metric to display. Repeat the flag for several metrics.
• --stat <name>: statistic to display per metric. One of min, median, max, average, p80, p90, p99, or eventCount.

The table shows median and eventCount by default, and merges them into one cell, such as 0.45s (150). The App version column includes the build number in parentheses. Update IDs are omitted from the table to keep it readable, but --json returns them as an array per version.

eas observe:metrics

Shows individual samples rather than aggregates. Use it to investigate outliers and to find the session behind a slow launch.

The metric is a positional argument. Omitting it prompts for a choice, and fails in non-interactive mode.

Command flags:

• --sort <order>: one of oldest (default), newest, slowest, or fastest.
• --limit <number>: samples per page. Defaults to 10, capped at 100.
• --after <cursor>: the endCursor from a previous run.
• --app-version <version>: filter by app version.
• --update-id <id>: filter by EAS Update ID.

When more results are available, the command prints the flag needed to fetch the next page. JSON output adds sessionId, easClientId, and any custom params attached to the sample.

eas observe:routes

Shows navigation metrics grouped by route name, with a separate section per platform. Use it to find the screens that are slowest to reach.

Command flags:

• --metric <name>: one of nav_cold_ttr, nav_warm_ttr, or nav_tti. Repeat the flag for several metrics. Defaults to all three.
• --stat <name>: one of median, p90, or count.
• --route-name <name>: filter by route name. Repeat the flag for several routes.
• --app-version <version> and --build-number <number>: filter to one release.
• --update-id <id>: filter by EAS Update ID.
• --limit <number>: routes per page. Defaults to 50, capped at 200.
• --after <cursor>: the endCursor from a previous run.

Route names are patterns, such as /(tabs)/sessions/[sessionId], so distinct parameter values group together. Each platform paginates separately, so the next-page hint names the platform it applies to.

eas observe:session

Shows every metric and log event recorded during one session, in order. Use it after observe:metrics surfaces a slow sample, to see what else happened during that launch.

The session ID is a positional argument. Omitting it in interactive mode prompts you to pick from a list of candidate sessions. In non-interactive mode, including under --json, the session ID is required. Session IDs are also included in the --json output of observe:metrics and observe:events.

Command flags:

• --event-name <name>: the metric or user-defined event used to build the candidate list, such as tti or onboarding.completed.
• --sort <order>: orders the candidate events. One of slowest, fastest, newest, or oldest.

The flags that build the candidate list — --event-name, --sort, --days, --start, and --end — describe how to find a session, so they cannot be combined with a session ID. Pass the ID on its own to inspect a session you already have.

eas observe:events

Shows user-defined events logged with Observe.logEvent, as well as events emitted by the SDK and its integrations, such as expo.memory.warning and expo-image.oversized. With no arguments, it lists event names and their counts.

Command flags:

• --all-events: list every event instead of the name summary. Cannot be combined with an event name.
• --session-id <id>: filter to one session. For the full timeline, including metrics, use observe:session.
• --app-version <version>: filter by app version.
• --update-id <id>: filter by EAS Update ID.
• --limit <number> and --after <cursor>: paginate the results.

Querying a name that has no events prints the available names for the same time range, which makes typos easy to spot.

eas observe:versions

Lists the app versions in the field with their build numbers, update IDs, and event counts. Use it to find the identifiers that the other commands filter by.

The table shows app version, first seen, events, users, builds, and updates. JSON output returns the full hierarchy, with EAS Build and update details nested under each version.

Common workflows

Compare the current release against the previous one:

Find and investigate the slowest launches:

Check which screens are slowest to become interactive:

Check how over-the-air updates download in the field:

Gate a script or CI job on a metric: