[Vadman97]

Demo (test transcript): team-o11y:artifacts/sf4t1/demo-test-output.txt

What and why

The React Native SDK only put context keys on the LD.identify log and LD.track spans, so you couldn't tie other RN rows to a user or device. Now every identify stores the context keys, and they're stamped as context.contextKeys.<kind> (same names as web) on all logs, errors, spans (manual, track, flag, network, console.error) and metrics. This uses the existing shared-attribute processors.

• InstrumentationManager.setContextKeyAttributes replaces the stored keys on each call and fills getContextKeyAttributes().
• An internal _LDObserve._setContextKeys call is buffered before init and goes through ObservabilityClient.setContextKeys.
• afterIdentify sets the keys first, whatever the identify status, as web does. The LD.identify log keeps its bare user/device fields for back-compat.

Testing

• yarn workspace @launchdarkly/observability-react-native test: 18 files, 155 tests pass, including the new src/client/ContextKeys.test.ts (12 tests). They cover log, error, console.error, manual/track/flag/XHR spans, all metric kinds, multi-context, re-identify replacing keys, a non-completed identify, and an identify before init being buffered.
• yarn workspace @launchdarkly/observability-react-native typegen: passes.

Open questions

• Keys are also set when identify returns error/shed, to match web. A later identify overwrites them. Should RN only set them on completed?
• Network spans are tested through XHR, not fetch: jsdom has no fetch. Both go through the same span processor.

🤖 Generated with Claude Code

---

Opened by termag ask for vkorolik: an agent run on sf4t1-10292244/observability-sdk, thread cmvu4tz271hq2f7cqjodjfd0a.

The run worked in its own clone of the committed code. termag checked its commits before pushing (descends from the base, no protected paths, authored by the asker, no secret-shaped strings); review the change as you would any other before marking it ready.

Follow-ups on the thread push here: termag ask sf4t1-10292244/observability-sdk --thread cmvu4tz271hq2f7cqjodjfd0a "…".

_{Opened from termag}

---

Note

Overview
React Native observability now attaches LaunchDarkly evaluation context to every signal, not only LD.identify / track paths. On each afterIdentify, the plugin pushes context keys through _LDObserve._setContextKeys (buffered before init), which lands in InstrumentationManager.setContextKeyAttributes and replaces the stored snapshot as context.contextKeys.<kind>—aligned with the web SDK naming.

Those keys merge with global attributes in getSharedAttributes(), so logs, errors, manual/track/flag/network spans, and metrics pick them up via the existing global-attribute processors. Keys are updated on every identify (including non-completed statuses, like web); re-identify replaces the set. The LD.identify log still includes bare user/device fields for back-compat.

Adds ContextKeys.test.ts (store/replace, metrics, plugin identify, multi-context, pre-init buffering, re-identify, failed identify).

^{Reviewed by Cursor Bugbot for commit a770a61. Bugbot is set up for automated code reviews on this repo. Configure here.}

[Vadman97]

bugbot run

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit a770a61. Configure here.}