meta-wearables-debugging-observability

Start by connecting to the local DAT Inspector debug server and checking get_connection_status, get_sdk_state, and get_dat_readiness in that order. Connection failure, SDK failure, and device-link failure are distinct states, so identify the first failing transition before attempting recovery.

Discover and connect to the local debug server, establish readiness with get_dat_readiness, then reproduce one narrow operation while capturing wait_for_events with the narrowest category. Use get_errors or get_event_digest for the digest and export_diagnostic_bundle for a redacted handoff.

Yes, the workflow covers both platforms with platform-specific anchors such as Wearables.configure() on iOS and Wearables.initialize(context) on Android. Each platform uses its own typed results and errors, and symbols should not be translated across platforms.

Remove credentials, client and access tokens, raw image, video, and audio, transcripts, personal data, precise device identifiers, and account emails. Keep state names, timestamps, SDK revisions, error categories, and counters needed for diagnosis, and replace identifiers with stable local labels.

A generic not connected message hides the actual first failing state, which could be registration, permission, device eligibility, link, session, or transport. Walk the boundary in order using readiness and companion-boundary diagnosis tools to observe the exact failing transition instead of guessing.

No, diagnosis is strictly read-only and must not mutate Meta AI, glasses, account, permission, registration, project, or release-channel state unless the user separately authorizes that operation. The workflow treats debugging as observation, not intervention.