Release 12-29-2025

[eosho]

This pull request introduces a new data visualization feature that allows users to generate charts and graphs from query results using natural language. The implementation leverages Azure Container Apps Dynamic Sessions for secure, isolated code execution of LLM-generated Python (matplotlib) code. The changes span documentation, configuration, environment setup, core code, and dependency management.

Key highlights:

• Adds support for natural language-driven data visualization by enabling a code interpreter that securely executes charting code in Azure.
• Updates documentation and configuration files to guide users on enabling and using the visualization feature.
• Introduces new configuration schema and validation for the code interpreter.
• Updates dependencies to include required packages for visualization and Azure integration.

---

Data Visualization Feature

• Added support for natural language-driven chart and graph generation using a code interpreter that runs LLM-generated matplotlib code in Azure Container Apps Dynamic Sessions. This includes intent detection, secure execution, and result embedding. [1] [2] [3] [4]
• Introduced new documentation (docs/VISUALIZATION.md) detailing setup, configuration, and architecture for the visualization feature.
• Updated README.md and configuration docs to describe the visualization feature, usage examples, and setup instructions. [1] [2]

Configuration and Environment

• Added code_interpreter configuration section to agent YAML files and schema, including validation for required Azure endpoint. Updated example config and schema to support the new feature. [1] [2] [3] [4] [5] [6] [7] [8]
• Added AZURE_SESSIONS_POOL_ENDPOINT environment variable to .env.example for configuring the Azure session pool endpoint.

Dependency Management

• Added new dependencies for visualization and Azure integration: langchain-azure-dynamic-sessions, matplotlib, and langchain-experimental.

---

These changes collectively enable secure, flexible, and easy-to-use data visualization capabilities for end users, while providing clear setup and configuration guidance.