Question

To speed up a slow analytics query, an AI agent proposed and wrote this migration adding several indexes to a high-write `events` table (≈30k inserts/sec). Read latency on the dashboard improved, but a week later on-call is paged for rising write latency and replication lag.

1CREATE INDEX idx_events_user   ON events (user_id);2CREATE INDEX idx_events_type   ON events (event_type);3CREATE INDEX idx_events_region ON events (region);4CREATE INDEX idx_events_status ON events (status);5CREATE INDEX idx_events_source ON events (source);

Critique this both for the change mechanics and the design.