Fujitsu Social Digital Twin MCP Facilitator

This integration layer connects the advanced Fujitsu Digital Rehearsal API with the Model Context Protocol (MCP). It translates high-level conversational instructions from LLMs into actionable calls against Fujitsu's simulation environment.

Core Functionality

The Fujitsu Digital Twin technology models the dynamics of populated areas, encompassing individual human behavior and large-scale economic activities based on empirical data. Its central feature, "Digital Rehearsal," permits the pre-verification of policy impacts by executing virtual trials of social and traffic interventions.

This MCP gateway bridges the linguistic gap, empowering users, via an LLM, to initiate complex simulations, derive analytical insights from outcomes, and manage persistent simulation assets using plain speech.

Capabilities Summary

• Cataloging and presenting available simulation setups.
• Initiating simulation execution runs.
• Fetching and interpreting simulation output metrics.
• Administration of stored simulation artifacts.
• Specialized analysis for urban mobility simulations.
• Comparative evaluation between distinct simulated conditions.
• Automatic blueprint generation for simulations based on textual descriptions.

Setup Requirements

• Operating environment requires Python version 3.13 or newer.
• Valid authentication credentials (API Key) for the Fujitsu API Gateway.
• An MCP-compliant interface client (e.g., Claude Desktop).

Deployment Procedure

Installation via Smithery (Automated)

Utilize the Smithery CLI for rapid deployment directly to clients like Claude Desktop:

bash npx -y @smithery/cli install @3a3/fujitsu-sdt-mcp --client claude

Manual Source Setup

1. Acquire Codebase bash git clone https://github.com/3a3/fujitsu-sdt-mcp.git cd fujitsu-sdt-mcp

2. Environment Configuration (Recommended: uv dependency manager) Install uv first: bash pip install uv # Or use curl for Linux/macOS curl -sSf https://astral.sh/uv/install.sh | sh

   Then, establish and populate the environment: bash uv venv

   Activation (Windows):

   .venv\Scripts\activate

   Activation (Unix/MacOS):

   source .venv/bin/activate

   Dependency resolution:

   uv pip install -r requirements.txt

   Alternative: Execute the provided setup script: bash chmod +x setup.sh ./setup.sh

3. Credential Configuration Credentials must be exposed via environment variables or a .env file.

   Environment Variables (Unix/MacOS Example): bash export FUJITSU_API_BASE_URL=https://apigateway.research.global.fujitsu.com/sdtp export FUJITSU_API_KEY=your_secret_key

   .env File Content:

   FUJITSU_API_BASE_URL=https://apigateway.research.global.fujitsu.com/sdtp FUJITSU_API_KEY=your_secret_key

Operational Guidance

Launching the MCP Backend Service

The server exposes its capabilities over standard input/output streams:

bash python -m fujitsu_sdt_mcp

Direct Client Testing

A supplementary utility script allows for immediate, command-line interaction testing:

bash python client.py

Integration with LLM Interfaces (e.g., Claude Desktop)

1. Access Claude Desktop configuration settings.
2. Locate and modify the relevant configuration file.
3. Incorporate the following directives:

"command": "uvx", "args": [ "fujitsu-sdt-mcp" ], "env": { "FUJITSU_API_BASE_URL": "https://apigateway.research.global.fujitsu.com/sdtp", "FUJITSU_API_KEY": "your_secret_key" }

1. Restart the client application.

Available Interfaces (Resources & Tools)

Data Access Endpoints (Resources)

• resource://digital_rehearsal_overview: Summary of the Digital Rehearsal technology suite.
• resource://simulation_metrics_explanation: Glossary defining output metrics.
• resource://scenario_examples: Repository of pre-built scenario blueprints.

Actionable Functions (Tools)

Tool Name	Purpose
list_simulations	Fetch registered simulation identifiers.
start_simulation	Initiate a new simulation run.
get_simulation_result	Retrieve finalized simulation outcomes.
get_metrics	Extract quantifiable performance indicators.
list_simdata	Catalog available input datasets.
get_simdata	Acquire specific input data sets.
analyze_traffic_simulation	Deep dive analysis on mobility runs.
compare_scenarios	Perform delta analysis between two runs.
create_natural_language_simulation_config	Synthesize configuration parameters from descriptive text.

Interaction Blueprint

Scenario: Optimizing Tokyo Traffic Flow

User Prompt: "I need to model traffic flow adjustments for Tokyo during peak morning hours."

LLM Response Path:

1. Configuration Generation: Translate request to parameters. [Tool execution: create_natural_language_simulation_config {"description": "I want to run a traffic optimization simulation during morning rush hour in Tokyo"}] (Generates necessary JSON configuration.)
2. Data Validation: Identify necessary input datasets. [Tool execution: list_simdata]
3. Execution: Launch the rehearsal. [Tool execution: start_simulation {"simdataId": "validated_tokyo_data_id"}] (Receives temporary ID: "sim_12345".)
4. Post-Simulation Analysis: Once complete, run evaluation. [Tool execution: analyze_traffic_simulation {"simulationId": "sim_12345", "region": "Tokyo", "timeRange": "morning rush hour", "scenario": "traffic optimization"}]

Analysis Highlights: (Example Output) * Observed travel time: 42 minutes (baseline). * Congestion hotspots identified on key arterial routes. * Recommendations provided for signal timing adjustments.

Governance and Support

Issues and feature solicitations should be submitted via the project's GitHub repository. Contributions through pull requests are welcomed.

Licensing

Distributed under the MIT License (refer to LICENSE file).

Credits

• Fujitsu Corporation - For underpinning Social Digital Twin methodology.
• Model Context Protocol (MCP) Community - For defining the integration standard.

---

WIKIPEDIA CONTEXT NOTE: Cloud computing, as defined by ISO, is an architecture enabling network-accessible, scalable, and elastic resource provisioning with self-management capabilities.