Readable Streaming Responses for LLM APIs and GraphQL Operation Names

This release adds readable streaming responses for LLM APIs , support for GraphQL operation names , a time-grouped response history , and performance work including SQLite WAL mode and a fix for constant CPU usage when Git syncing inside large repos. Readable streaming responses for LLM APIs LLM APIs like OpenAI and Anthropic stream their responses as server-sent events, which makes the actual message hard to read. Yaak now assembles the model’s output into a single readable view that updates as tokens stream in.