Convex Agents Messages

To save user messages to preserve conversation history, you use thread-based storage with typed Message schemas to reliably retain user prompts and context for later retrieval.

You retrieve messages by thread to render chat history, utilizing UIMessages with UI-related fields and utilities for optimistic updates to support rich rendering in the chat interface.

You filter tool messages from history by applying built-in utilities that remove tool-specific entries from the thread, keeping only relevant context to display in the chat UI.

Yes, this approach supports optimistic updates for message management by providing utilities that instantly render user prompts in the UI before backend generation completes.

Yes, you can use typed Message schemas for thread-based storage to enforce data consistency when saving and retrieving chat history for UI rendering.

The best way to render agent chat history with UIMessages is to retrieve thread-based messages and map their typed schemas to UI-related fields, applying optimistic updates for smooth display.