Trade Surveillance

You deploy and run the Trade Surveillance MCP Server to automate trade surveillance workflows by parsing inquiries, searching SQL configs and Java code, executing reports, and generating responses. It integrates with your existing code repositories and configurations to streamline support operations using keyword-based search and automation.

How to use

Open a compatible MCP client and connect to the Trade Surveillance MCP Server. You will be able to paste user inquiry emails, and the server will parse key details, search your SQL configs and Java code by meaningful metadata, run Java reports, and generate a cohesive response summary for the user.

How to install

Prerequisites: Python 3.10 or higher and the uv package manager (recommended) or pip.

# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install the MCP server with uv
uv pip install -e .

Or install with pip directly (no uv):

pip install -e .

Configuration and setup examples

Configure the MCP client integration to connect to the Trade Surveillance MCP Server using a local stdio setup. The exact command and arguments run the server from your environment, allowing you to start the server within your development workspace.

{
  "mcp.servers": {
    "trade_surveillance": {
      "type": "stdio",
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/mcp_test_2",
        "run",
        "trade-surveillance-mcp"
      ]
    }
  }
}

Alternatively, configure the same server in Claude Desktop so you can start it from that environment.

{
  "mcpServers": {
    "trade_surveillance": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/mcp_test_2",
        "run",
        "trade-surveillance-mcp"
      ]
    }
  }
}

Run and test locally

Start the MCP server locally using the recommended approach. You can run it directly or via Python in development mode.

# Run the server directly with uv
uv run trade-surveillance-mcp
`

Or start it via Python if you prefer the module entry point.

python -m trade_surveillance_mcp.server

Notes on usage and tools

This MCP server uses metadata-based searches to locate relevant SQL configs and Java code by keywords rather than by file paths. You will interact with a set of tools designed to parse inquiries, search metadata-labeled files, run Java reports, rebuild the metadata index when new files are added, and generate final user-friendly summaries.

Troubleshooting

If the server does not appear in your MCP client, verify the absolute path in the configuration is correct and that uv is installed and available in your PATH.

Ensure you are running Python 3.10+ and that your environment has access to Java for executing Java reports.

Additional notes

Customize the server to your repository layout by updating search paths and adding your own email parsing, SQL, and Java search logic. After adding new files or updating annotations, rebuild the metadata index to keep searches accurate.

Available tools

parse_email_inquiry

Parses user inquiry emails to extract inquiry type, trade IDs, time periods, and priority.

search_sql_configs

Searches SQL configuration files by keywords using metadata annotations instead of file paths.

search_java_code

Finds Java classes and methods by keywords using javadoc annotations rather than file locations.

execute_java_report

Runs Java processes to generate required reports from the identified configurations.

rebuild_metadata_index

Rebuilds the search index by scanning all annotated SQL configs and Java files.

generate_response_summary

Generates a comprehensive summary response for the user based on search results and reports.