Add endpoint to clear cached multi-agent responses in Redis

Update OneDrive folder name for HDF submissions to "Health Declaration Form"

Add public HDF submission endpoint with fixed account handling; implement upload size checks and secret validation

Refactor OneDrive upload to use fixed clinic accounts based on save type; update documentation for target account handling

Add startup diagnostics for OneDrive configuration: log presence of AZURE_CLIENT_ID, AZURE_CLIENT_SECRET, and count of parsed tokens; warn on malformed ONEDRIVE_REFRESH_TOKENS

Refactor OneDrive integration for per-user uploads: implement user-specific OAuth2 token handling and update documentation for refresh token setup

Add OneDrive integration: implement upload functionality for clinic charts/reports and create OAuth2 token minting script

Enhance Gmail integration: add endpoints for marking emails as read/unread and drafting replies; refactor inbox fetching to support unread-only filtering and improve email handling

Add support for managing ignored senders in Gmail integration; implement unignore functionality and enhance attachment handling

Add ignore sender functionality to Gmail integration; implement endpoints for managing ignored senders and update triage logic

Enhance Gmail integration to support email trashing; add trash candidates in responses and update documentation

Refactor Gmail integration to use REST API with OAuth2

Refactor Gmail agent to improve error handling and add timeouts for IMAP and LLM calls; remove unused code and enhance logging for better diagnostics

Add Gmail integration for inbox triage and reply drafting; implement shared mailbox access and user-defined writing style

Update HISTORY_PERIOD to 1 year and optimize model prediction call in stock LSTM forecasting

Add computer vision models for image classification and object detection; update Dockerfile and requirements

Remove training script for stock LSTM model and update dependencies; add support for new ticker D05.SI

Add LSTM training script and update requirements for model artifacts

Revert main to a344aea state

Update requirements and improve stock LSTM demo handling; adjust cache TTL values

Add image classification and object detection routes; include MobileNet and Faster R-CNN models

Refactor caching logic to improve cacheability determination based on conversation context; ensure session turns are persisted even on cache hits.

Update RAG_SYSTEM_PROMPT to restrict output format; disallow Markdown or HTML tables for tabular data presentation

Implement data-source access control for SQL agent; add caching for user access profiles and enhance query processing with access validation

Refactor multi-agent chat handling with improved caching, logging, and session management; enhance database schema for feedback logging

Enhance message routing with LLM integration, add image request override, and improve routing logic

Refactor agents to use asynchronous clients, enhance SQL execution with connection pooling, and improve rate-limiting logic

Enhance multi-agent routing with improved logging, timeout handling, and input sanitization

Implement admin user approval flow and enhance user status verification

Add history-aware query rewriting and improve follow-up detection in RAGAgent

Add caching to embedding function to reduce latency during RAG queries

Add history-aware query rewriting and groundedness scoring to RAG agent

Increase expiration time for public document URLs to 3600 seconds and enhance URL handling for website-sourced documents

Refactor RAG chunking and retrieval logic; add match threshold and metadata filters for enhanced document retrieval

Add cross-encoder reranking support for improved answer accuracy and update requirements

Refactor token decoding logic to enhance support for asymmetric and symmetric JWTs

Refactor MessageRouter to replace sales and inventory keywords with database-related keywords

Add image generation functionality with DeepSeek and Hugging Face integration

Remove infographic generation functionality and related imports from agents orchestration

Add Docker support with docker-compose for FastAPI and Redis setup

Refactor SQL queries to improve aliasing for clarity and consistency

Refactor SQL queries to improve readability and consistency in naming conventions

Increase max_tokens limit for SQLAgent to enhance response capacity

Refactor SQL example inputs to use consistent aliasing for clarity and readability

Revert "Clarify SQL example inputs in system prompt to enhance user understanding and ensure consistency in query structure"

Clarify SQL example inputs in system prompt to enhance user understanding and ensure consistency in query structure

Revise SQL response prompts to clarify the order of results presentation, emphasizing business-friendly summaries followed by markdown tables.

Strip LLM preamble from SQL statements by identifying the start of the SQL query

Enhance SQL response formatting by requiring markdown tables for query results in prompts

Refactor rate limiting implementation by creating a separate limiter module and applying limits to various routes