This project provides Model Context Protocol (MCP) servers for interacting with the Apache Fineract API, enabling AI agents to access financial data and operations. Implementations are available in Python, Java (Quarkus), and Node.js.

Use the MCP Inspector to test and debug your server:

npx @modelcontextprotocol/inspector

This starts a local web UI to connect to your MCP server via STDIO or SSE.

Prerequisites: Python 3.8+, flask, mcp.server.fastmcp

Steps:

1. Install dependencies:

   pip install mcp[cli] uv flask

   Note for zsh users: If you're using zsh, be sure to quote extras to avoid shell expansion errors:

   pip install 'mcp[cli]' uv flask

2. Run the server:

   mcp dev app.py

Prerequisites: JDK 17+, Maven

Steps:

1. Configure environment variables in your shell or IDE:

   export MIFOSX_BASE_URL="https://your-fineract-instance"
   export MIFOSX_BASIC_AUTH_TOKEN="your_api_token"
   export MIFOS_TENANT_ID="default"

2. Run via JBang (for quick execution):

   jbang --quiet org.mifos.community.ai:mcp-server:1.0.0-SNAPSHOT:runner

3. (Optional) Build a native executable:

   ./mvnw package -Dnative
   ./target/mcp-server-1.0.0-SNAPSHOT-runner

Prerequisites: Node.js 16+, npm

Steps:

1. Install dependencies:

   cd nodejs && npm install

2. Configure environment variables in .env:

   cp .env.example .env

3. Run the server:

   npm run dev

4. Test with the built-in inspect script:

   npm run inspect

All implementations require the following environment variables:

Note: Java uses MIFOSX_ prefixed variables (e.g., MIFOSX_BASE_URL).

The MCP server exposes these resources:

• fineract://clients
  List all clients
• fineract://clients/{clientId}
  Get details for a specific client
• fineract://loans
  List all loans
• fineract://loans/{loanId}
  Get details for a specific loan

• search_clients
  Search clients by name/attributes
• create_client
  Create a new client (Node.js/Python only)
• update_loan_status
  Update loan status (Java/Python only)

For Java (Quarkus), create a native executable:

./mvnw package -Dnative -Dquarkus.native.container-build=true
./target/mcp-server-1.0.0-SNAPSHOT-runner

1. Start your MCP server (Python/Java/Node.js).
2. Run the inspector:

   npx @modelcontextprotocol/inspector

3. Connect to the server using the STDIO transport.

• Python: Modify python/app.py and server.js for new resources.
• Java: Extend src/main/java/org/mifos/community/ai/... for new endpoints.
• Node.js: Update nodejs/src/server.js and add Zod schemas for validation.

• Apache Fineract Community: https://community.apache.org/
• MCP Specification: https://modelcontextprotocol.org

• Java/Quarkus: Quarkus MCP Guide
• Node.js: Use npm run inspect for live reloading
• Python: Run with python app.py and configure .env

• Standardized API access via fineract:// URIs
• MCP-compliant with STDIO/SSE transports
• Environment-agnostic configuration