Greptile Summary

This PR introduces selective feature flag evaluation optimization to the PostHog Python SDK. The key change adds a flag_keys_to_evaluate parameter to all feature flag evaluation methods (get_flags_decision(), get_all_flags(), get_all_flags_and_payloads(), get_feature_variants(), get_feature_payloads(), and get_feature_flags_and_payloads()). This parameter allows clients to specify exactly which feature flags they want to evaluate instead of evaluating all available flags.

The implementation adds a flag_keys_filter option to the send_feature_flags parameter in capture events, enabling selective flag evaluation during event tracking. The optimization works at two levels: for remote evaluation, it passes the specified flag keys to the /flags endpoint to reduce server-side processing and network overhead; for local evaluation, it filters the flag processing to only evaluate the requested flags.

The change also optimizes single flag evaluation by making _get_feature_flag_details_from_decide request only the specific flag being evaluated rather than all flags. This addresses performance concerns for applications with many feature flags where only a subset needs evaluation. The implementation maintains backward compatibility by making the parameter optional with a default of None, preserving existing behavior when not specified.

Confidence score: 4/5

• This PR is generally safe to merge but requires careful review of the parameter passing logic
• Score reflects well-structured implementation with comprehensive tests, but the complex parameter threading across multiple methods introduces potential for subtle bugs
• Pay close attention to posthog/client.py for proper parameter handling and the test files for coverage validation

_{6 files reviewed, no comments}

_{Edit Code Review Bot Settings | Greptile}