The profiles table previously lacked an index on the user_id column, causing full table scans during user lookups. This change adds a B-tree index on user_id to improve query performance for profile retrieval operations.
The profiles table previously lacked an index on the user_id column, causing full table scans during user profile retrieval. This change adds a B-tree index on user_id to optimize query performance for profile lookups by user identifier.
This commit introduces a full-featured agent architecture including persistent memory storage using vector embeddings, multi-step planning capabilities with dynamic replanning, and an extensible tool registry supporting custom function definitions. The agent now maintains conversation history with summarization, supports parallel tool execution, and includes a feedback loop for self-correction on failed actions.
The profiles table previously lacked an index on the user_id column, causing full table scans during user lookups. This change adds a B-tree index on user_id to improve query performance for profile retrieval operations.
Implement avatar management with image upload, client-side cropping, and server-side storage. Users can now upload profile pictures from their local filesystem, crop them to a square aspect ratio using a drag-and-resize interface, and save the result. The backend stores cropped images in a dedicated avatars directory and updates the user record with the new avatar path. Includes validation for file type and size limits, plus fallback to default avatar on error.