MCP Integration in Streamlit Dashboard

Overview

The MCP (Model Context Protocol) server has been integrated directly into the Streamlit dashboard, allowing you to fetch data from various sources without leaving the app.

How to Use

1. Access the Data Fetching View

Open the Streamlit dashboard
In the sidebar, select “📥 Fetch Data Sources” from the Analysis View options
You’ll see a new section for fetching data from external sources

2. Available Data Sources

Anthropic EconomicIndex

Source: Hugging Face dataset
Usage: Select “Anthropic EconomicIndex”, enter release version, click “Fetch Data”
Returns: Economic indicators related to AI impacts
Requires: pip install datasets (if MCP not available)

Stanford AI Index

Source: Stanford HAI AI Index Report 2025
Usage: Select “Stanford AI Index”, choose metric type, click “Fetch Data”
Returns: AI investment, adoption, and performance metrics
Metrics Available:
- Investment (US: $109.1B, China: $9.3B, UK: $4.5B)
- Adoption (78% of organizations using AI in 2024)
- Performance metrics

World Bank Indicator

Source: World Bank API
Usage: Select “World Bank Indicator”, enter indicator code, country, and year range
Common Indicators:
- SL.EMP.ICTI.ZS - ICT Services Employment
- SL.EMP.ICTM.ZS - ICT Manufacturing Employment
- IT.NET.USER.ZS - Internet Users
Returns: Time series data with statistics

ITU ICT Data

Source: ITU via World Bank API
Usage: Select “ITU ICT Data”, choose indicator (internet/mobile/broadband), enter country
Returns: ICT penetration and usage data

View Source Information

Usage: Select “View Source Information”, choose a source
Returns: Information about data sources, URLs, and access methods

Features

Automatic Fallback

If MCP SDK is not installed, the app uses direct API access
World Bank data uses direct API calls (faster)
Anthropic data can use Hugging Face datasets library directly
Stanford data shows key metrics even without MCP

Data Display

Fetched data is displayed in interactive tables
Statistics are shown as metrics
Data can be viewed, analyzed, and exported

Error Handling

Clear error messages if data fetch fails
Installation instructions for missing dependencies
Graceful degradation if services are unavailable

Code Structure

Files

mcp_client.py - Simplified MCP client wrapper
- MCPDataFetcher class
- Methods for each data source
- Fallback mechanisms
app.py - Updated Streamlit app
- New “Fetch Data Sources” view
- UI components for data fetching
- Integration with MCP client

Usage in Code

from mcp_client import data_fetcher

# Fetch Anthropic data
result = data_fetcher.fetch_anthropic_data("release_2025_09_15")

# Fetch Stanford AI Index
result = data_fetcher.fetch_stanford_ai_index("investment")

# Fetch World Bank indicator
result = data_fetcher.fetch_world_bank_indicator(
    "SL.EMP.ICTI.ZS", 
    "USA", 
    2015, 
    2024
)

Installation

Option 1: With MCP (Full Features)

pip install mcp

Option 2: Without MCP (Fallback Mode)

# For Anthropic data
pip install datasets

# World Bank and ITU work without MCP (direct API)

Example Workflow

Start Dashboard:
```
streamlit run app.py
```
Navigate to Fetch Data:
- Select “📥 Fetch Data Sources” from sidebar
Fetch Data:
- Choose a data source
- Configure parameters
- Click “Fetch Data”
View Results:
- Data appears in tables
- Statistics shown as metrics
- Can be analyzed further
Integrate Data:
- Use fetched data for analysis
- Compare with existing dashboard data
- Export if needed

Troubleshooting

MCP Not Available

The app will use direct API access
World Bank data works without MCP
Some features may be limited

Import Errors

# Install missing dependencies
pip install mcp datasets requests pandas

Data Fetch Failures

Check internet connection
Verify API endpoints are accessible
Check error messages for specific issues

Next Steps

Test Data Fetching: Try fetching from each source
Integrate Data: Use fetched data in dashboard analysis
Extend Functionality: Add more data sources as needed
Save Data: Optionally save fetched data to database

Support

For issues: - Check mcp_server/README.md for MCP server details - Review DATA_SOURCES.md for source information - See INTEGRATION_GUIDE.md for integration steps

# MCP Integration in Streamlit Dashboard ## Overview The MCP (Model Context Protocol) server has been integrated directly into the Streamlit dashboard, allowing you to fetch data from various sources without leaving the app. ## How to Use ### 1. Access the Data Fetching View 1. Open the Streamlit dashboard 2. In the sidebar, select **"📥 Fetch Data Sources"** from the Analysis View options 3. You'll see a new section for fetching data from external sources ### 2. Available Data Sources #### Anthropic EconomicIndex - **Source:** Hugging Face dataset - **Usage:** Select "Anthropic EconomicIndex", enter release version, click "Fetch Data" - **Returns:** Economic indicators related to AI impacts - **Requires:** `pip install datasets` (if MCP not available) #### Stanford AI Index - **Source:** Stanford HAI AI Index Report 2025 - **Usage:** Select "Stanford AI Index", choose metric type, click "Fetch Data" - **Returns:** AI investment, adoption, and performance metrics - **Metrics Available:** - Investment (US: $109.1B, China: $9.3B, UK: $4.5B) - Adoption (78% of organizations using AI in 2024) - Performance metrics #### World Bank Indicator - **Source:** World Bank API - **Usage:** Select "World Bank Indicator", enter indicator code, country, and year range - **Common Indicators:** - `SL.EMP.ICTI.ZS` - ICT Services Employment - `SL.EMP.ICTM.ZS` - ICT Manufacturing Employment - `IT.NET.USER.ZS` - Internet Users - **Returns:** Time series data with statistics #### ITU ICT Data - **Source:** ITU via World Bank API - **Usage:** Select "ITU ICT Data", choose indicator (internet/mobile/broadband), enter country - **Returns:** ICT penetration and usage data #### View Source Information - **Usage:** Select "View Source Information", choose a source - **Returns:** Information about data sources, URLs, and access methods ## Features ### Automatic Fallback - If MCP SDK is not installed, the app uses direct API access - World Bank data uses direct API calls (faster) - Anthropic data can use Hugging Face datasets library directly - Stanford data shows key metrics even without MCP ### Data Display - Fetched data is displayed in interactive tables - Statistics are shown as metrics - Data can be viewed, analyzed, and exported ### Error Handling - Clear error messages if data fetch fails - Installation instructions for missing dependencies - Graceful degradation if services are unavailable ## Code Structure ### Files 1. **`mcp_client.py`** - Simplified MCP client wrapper - `MCPDataFetcher` class - Methods for each data source - Fallback mechanisms 2. **`app.py`** - Updated Streamlit app - New "Fetch Data Sources" view - UI components for data fetching - Integration with MCP client ### Usage in Code ```python from mcp_client import data_fetcher # Fetch Anthropic data result = data_fetcher.fetch_anthropic_data("release_2025_09_15") # Fetch Stanford AI Index result = data_fetcher.fetch_stanford_ai_index("investment") # Fetch World Bank indicator result = data_fetcher.fetch_world_bank_indicator( "SL.EMP.ICTI.ZS", "USA", 2015, 2024 ) ``` ## Installation ### Option 1: With MCP (Full Features) ```bash pip install mcp ``` ### Option 2: Without MCP (Fallback Mode) ```bash # For Anthropic data pip install datasets # World Bank and ITU work without MCP (direct API) ``` ## Example Workflow 1. **Start Dashboard:** ```bash streamlit run app.py ``` 2. **Navigate to Fetch Data:** - Select "📥 Fetch Data Sources" from sidebar 3. **Fetch Data:** - Choose a data source - Configure parameters - Click "Fetch Data" 4. **View Results:** - Data appears in tables - Statistics shown as metrics - Can be analyzed further 5. **Integrate Data:** - Use fetched data for analysis - Compare with existing dashboard data - Export if needed ## Troubleshooting ### MCP Not Available - The app will use direct API access - World Bank data works without MCP - Some features may be limited ### Import Errors ```bash # Install missing dependencies pip install mcp datasets requests pandas ``` ### Data Fetch Failures - Check internet connection - Verify API endpoints are accessible - Check error messages for specific issues ## Next Steps 1. **Test Data Fetching:** Try fetching from each source 2. **Integrate Data:** Use fetched data in dashboard analysis 3. **Extend Functionality:** Add more data sources as needed 4. **Save Data:** Optionally save fetched data to database ## Support For issues: - Check `mcp_server/README.md` for MCP server details - Review `DATA_SOURCES.md` for source information - See `INTEGRATION_GUIDE.md` for integration steps