Release 12-29-2025

.env.example

@@ -87,3 +87,10 @@ BIGQUERY_DATASET=your-dataset
 BIGQUERY_LOCATION=US
 GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account.json
 BIGQUERY_CREDENTIALS_JSON=
+
+# =============================================================================
+# Visualization (Optional)
+# Requires Azure Container Apps Dynamic Sessions for secure code execution.
+# See docs/CONFIGURATION.md for setup instructions.
+# =============================================================================
+AZURE_SESSIONS_POOL_ENDPOINT=https://eastus.dynamicsessions.io/subscriptions/xxx/resourceGroups/xxx/sessionPools/xxx

README.md

@@ -28,6 +28,7 @@ Built on top of LangChain's [`SQLDatabase`](https://docs.langchain.com/oss/pytho
 - **Intent Detection**: Automatically routes queries to the correct data agent based on question context
 - **Multi-Turn Conversations**: Follow-up questions with context awareness (e.g., "What's the average?" after a query)
 - **SQL Validation**: Safe query execution with sqlglot-based validation across all dialects
+- **Data Visualization**: Generate charts and graphs from query results using natural language (e.g., "show me a bar chart")
 - **Configurable Agents**: YAML-based configuration for adding new data sources
 - **A2A Protocol**: Agent-to-Agent interoperability for integration with other A2A-compliant systems
 
@@ -49,6 +50,7 @@ Generates, validates, and executes SQL queries with retry logic.
 
 - [Database Setup](docs/DATABASE_SETUP.md)
 - [Configuration](docs/CONFIGURATION.md)
+- [Data Visualization](docs/VISUALIZATION.md)
 - [A2A Protocol](docs/A2A.md)
 
 ## Quick Start
@@ -163,6 +165,7 @@ data-agent chat -c adventure_works
 1. What are the total deposits by customer segment?
 2. Show me all high-severity fraud alerts from the past week
 3. Who are the top 5 customers by transaction volume?
+4. Show me a bar chart of transactions by type
 
 ```bash
 data-agent query "What are the total deposits by customer segment?" -c amex

docs/CONFIGURATION.md

@@ -35,6 +35,8 @@ data_agents:
       blocked_functions:
         - pg_sleep
         - pg_read_file
+    code_interpreter:
+      enabled: true
     system_prompt: |
       You are an SQL assistant...
       {schema_context}
@@ -54,6 +56,25 @@ data_agents:
         answer: "There are 1,234 users."
 ```
 
+## Code Interpreter (Data Visualization)
+
+Enable the code interpreter to generate charts and visualizations from query results. When enabled, the LLM can detect visualization intent (e.g., "show me a chart", "visualize", "plot") and generate matplotlib code to create charts.
+
+```yaml
+code_interpreter:
+  enabled: true
+  azure_sessions_endpoint: ${AZURE_SESSIONS_POOL_ENDPOINT}
+```
+
+| Setting | Description | Default |
+|---------|-------------|---------|
+| `enabled` | Enable/disable visualization generation | `false` |
+| `azure_sessions_endpoint` | Azure Container Apps session pool management endpoint URL | - |
+
+**Note:** Visualization requires Azure Container Apps Dynamic Sessions for secure, isolated code execution.
+
+See [VISUALIZATION.md](VISUALIZATION.md) for complete setup instructions, architecture details, and troubleshooting.
+
 ## SQL Validation
 
 Each data agent can configure SQL validation settings to control query safety:

docs/VISUALIZATION.md

@@ -0,0 +1,226 @@
+# Data Visualization
+
+This guide covers the code interpreter feature for generating charts and visualizations from query results.
+
+## Overview
+
+When enabled, the data agent can detect visualization intent in user queries (e.g., "show me a chart", "plot the data") and generate matplotlib code to create charts. The code runs in a secure, isolated environment using Azure Container Apps Dynamic Sessions.
+
+**Key features:**
+- Automatic detection of visualization requests
+- LLM-generated matplotlib code
+- Secure sandboxed execution with Hyper-V isolation
+- Native image capture (no file storage)
+- Support for bar charts, line charts, pie charts, scatter plots, and more
+
+## Requirements
+
+Visualization requires Azure Container Apps Dynamic Sessions. This provides:
+
+| Feature | Benefit |
+|---------|---------|
+| **Hyper-V isolation** | Each execution runs in a dedicated VM |
+| **Pre-installed packages** | NumPy, Pandas, Matplotlib ready to use |
+| **Native image capture** | `plt.show()` output captured automatically |
+| **Automatic cleanup** | Sessions terminate after idle timeout |
+| **No host access** | Code cannot access host filesystem or network |
+
+## Azure Setup
+
+### 1. Create a Container Apps Environment
+
+If you don't already have one:
+
+```bash
+az containerapp env create \
+    --name aca-env \
+    --resource-group rg-data-agent \
+    --location eastus
+```
+
+### 2. Create the Session Pool
+
+```bash
+az containerapp sessionpool create \
+    --name session-pool-viz \
+    --resource-group rg-data-agent \
+    --container-type PythonLTS \
+    --max-sessions 100 \
+    --cooldown-period 300 \
+    --location eastus
+```
+
+**Parameters:**
+- `--container-type PythonLTS`: Python runtime with common data science packages
+- `--max-sessions`: Maximum concurrent sessions
+- `--cooldown-period`: Seconds before idle session is terminated
+
+### 3. Get the Pool Management Endpoint
+
+```bash
+az containerapp sessionpool show \
+    --name session-pool-viz \
+    --resource-group rg-data-agent \
+    --query "properties.poolManagementEndpoint" -o tsv
+```
+
+This returns a URL like:
+```
+https://eastus.dynamicsessions.io/subscriptions/<sub>/resourceGroups/<rg>/sessionPools/<pool>
+```
+
+### 4. Assign the Executor Role
+
+Grant your identity permission to execute code in the session pool:
+
+```bash
+# Get your user ID
+USER_ID=$(az ad signed-in-user show --query id -o tsv)
+
+# Get the session pool resource ID
+POOL_ID=$(az containerapp sessionpool show \
+    --name session-pool-viz \
+    --resource-group rg-data-agent \
+    --query id -o tsv)
+
+# Assign the role
+az role assignment create \
+    --role "Azure ContainerApps Session Executor" \
+    --assignee $USER_ID \
+    --scope $POOL_ID
+```
+
+**Note:** For service principals or managed identities, replace `$USER_ID` with the appropriate object ID.
+
+### 5. Install the SDK
+
+```bash
+pip install langchain-azure-dynamic-sessions
+```
+
+Or add to your `pyproject.toml`:
+```toml
+dependencies = [
+    "langchain-azure-dynamic-sessions>=0.1.0",
+]
+```
+
+## Configuration
+
+### Environment Variable
+
+Set the pool endpoint:
+
+```bash
+export AZURE_SESSIONS_POOL_ENDPOINT="https://eastus.dynamicsessions.io/subscriptions/.../sessionPools/..."
+```
+
+Or in `.env`:
+```bash
+AZURE_SESSIONS_POOL_ENDPOINT=https://eastus.dynamicsessions.io/subscriptions/.../sessionPools/...
+```
+
+### YAML Configuration
+
+Enable visualization in your agent config:
+
+```yaml
+data_agents:
+  - name: "sales_agent"
+    # ... other config ...
+    code_interpreter:
+      enabled: true
+      azure_sessions_endpoint: ${AZURE_SESSIONS_POOL_ENDPOINT}
+```
+
+| Setting | Description | Default |
+|---------|-------------|---------|
+| `enabled` | Enable/disable visualization | `false` |
+| `azure_sessions_endpoint` | Session pool management endpoint URL | - |
+
+### System Prompt
+
+To enable visualization detection, include `visualization_requested` in your response format:
+
+```yaml
+system_prompt: |
+  You are a SQL expert for the sales database.
+
+  {schema_context}
+
+  ## Response Format
+
+  Provide your response as JSON with these fields:
+  - "thinking": Step-by-step reasoning about the query
+  - "sql_query": The generated SQL query
+  - "explanation": Brief explanation of what the query does
+  - "visualization_requested": Set to true if the user asks for a chart, graph, plot, or visualization
+```
+
+## How It Works
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│                        User Query                                │
+│            "Show me a bar chart of sales by region"             │
+└─────────────────────────────────────────┬───────────────────────┘
+                                          │
+                                          ▼
+┌─────────────────────────────────────────────────────────────────┐
+│                      SQL Generation LLM                          │
+│  Generates SQL + sets visualization_requested: true              │
+└─────────────────────────────────────────┬───────────────────────┘
+                                          │
+                                          ▼
+┌─────────────────────────────────────────────────────────────────┐
+│                     Database Query                               │
+│              Execute SQL, return result rows                     │
+└─────────────────────────────────────────┬───────────────────────┘
+                                          │
+                                          ▼
+┌─────────────────────────────────────────────────────────────────┐
+│                   Visualization LLM                              │
+│    Generates matplotlib code based on data + user question       │
+└─────────────────────────────────────────┬───────────────────────┘
+                                          │
+                                          ▼
+┌─────────────────────────────────────────────────────────────────┐
+│              Azure Container Apps Dynamic Sessions               │
+│   • Code executed in Hyper-V isolated container                 │
+│   • plt.show() output captured automatically                    │
+│   • Image returned as base64 PNG                                │
+└─────────────────────────────────────────┬───────────────────────┘
+                                          │
+                                          ▼
+┌─────────────────────────────────────────────────────────────────┐
+│                      Response                                    │
+│         Text explanation + embedded chart image                  │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+### Execution Flow
+
+1. **Intent Detection**: The SQL LLM sets `visualization_requested: true` when it detects chart/graph/plot intent
+2. **SQL Execution**: Query runs against the database, returning structured data
+3. **Code Generation**: A second LLM call generates matplotlib code tailored to the data and question
+4. **Sandboxed Execution**: Code runs in Azure Sessions with automatic image capture
+5. **Response Assembly**: Text response and chart image are combined for display
+
+## Example Queries
+
+These prompts trigger visualization:
+
+| Query | Chart Type |
+|-------|------------|
+| "Show me a bar chart of sales by region" | Bar chart |
+| "Visualize the top 10 customers by revenue" | Horizontal bar |
+| "Plot monthly revenue trends for 2024" | Line chart |
+| "Create a pie chart of transaction types" | Pie chart |
+| "Graph the distribution of order values" | Histogram |
+| "Compare Q1 vs Q2 performance" | Grouped bar |
+
+## Further Reading
+
+- [Azure Container Apps Dynamic Sessions](https://learn.microsoft.com/azure/container-apps/sessions)
+- [Session Pool Management](https://learn.microsoft.com/azure/container-apps/sessions-code-interpreter)
+- [LangChain Azure Dynamic Sessions](https://python.langchain.com/docs/integrations/tools/azure_dynamic_sessions)

pyproject.toml

@@ -22,15 +22,11 @@ dependencies = [
     "pyyaml>=6.0",
     "python-dotenv>=1.2.1",
     "databricks-sql-connector>=4.2.2",
-    "asyncpg>=0.29.0",
     "psycopg[binary]>=3.1.0",
-    "nest-asyncio>=1.6.0",
     "azure-cosmos>=4.7.0",
-    "aiohttp>=3.9.0",
     "structlog>=24.0.0",
     "typing-extensions>=4.12",
     "azure-identity>=1.25.1",
-    "azure-keyvault-secrets>=4.10.0",
     "langgraph-cli[inmem]>=0.4.10",
     "langgraph-api>=0.5.42",
     "pyodbc>=5.3.0",
@@ -41,14 +37,15 @@ dependencies = [
     "rich>=14.0.0",
     "chainlit>=2.0.0",
     "pandas>=2.0.0",
-    "tabulate>=0.9.0",
     "a2a-sdk[http-server]>=0.3.22",
     "httpx>=0.27.0",
     "sqlalchemy>=2.0.45",
     "sqlalchemy-bigquery>=1.16.0",
     "databricks-sqlalchemy>=2.0.8",
     "google-cloud-bigquery-storage>=2.36.0",
     "psycopg2>=2.9.11",
+    "langchain-azure-dynamic-sessions>=0.2.0",
+    "matplotlib>=3.10.8",
 ]
 
 [project.scripts]
@@ -72,15 +69,6 @@ dev = [
     "isort>=7.0.0",
 ]
 
-# Azure AI Foundry hosting dependencies
-foundry = [
-    "azure-ai-agentserver-langgraph>=0.1.0",
-    "azure-ai-projects>=1.0.0",
-    "azure-identity>=1.25.1",
-    "opentelemetry-api>=1.20.0",
-    "opentelemetry-sdk>=1.20.0",
-]
-
 [tool.setuptools.packages.find]
 where = ["src"]
 include = ["data_agent*"]

src/data_agent/agent.py

@@ -561,6 +561,9 @@ async def run(
             datasource_name=result.get("datasource_name", ""),
             rewritten_question=result.get("rewritten_question", ""),
             messages=result.get("messages", []),
+            visualization_image=result.get("visualization_image"),
+            visualization_code=result.get("visualization_code"),
+            visualization_error=result.get("visualization_error"),
         )
 
     def get_agent_names(self) -> list[str]: