Business Request Server

by dto-btn

148 downloads
Not rated
GitHub

About

The Business Request Server is a demo proof-of-concept MCP server designed for use in other PoCs. It provides MCP tools to search and filter business request records via a database connection (using pymssql).

Details

Author
dto-btn
Downloads
148
Categories
Other

- Provides search_business_requests tool with custom query filters
- Provides filter_results tool for post-search filtering
- Supports local development via mcp dev or python server.py
- Can be containerized with Docker
- Supports manual deployment to Azure Web Apps

Setting up with Highlight

This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:

  1. Download and install Highlight from highlightai.com/download
  2. Navigate to the plugins tab and select "Add Custom Plugin"
  3. Configure the plugin with the settings below
    Plugin Name Business Request Server
    Command (node, npx, python, etc.)

    Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.

  4. Enable "Start Automatically" if you want the plugin to start when Highlight launches

From the repository

Install dependencies with uv venv and uv pip install -e ., then run locally with mcp dev server.py or python server.py. Navigate to the URL shown to test the server. Use the search_business_requests tool with a JSON query filter and the filter_results tool to further refine results. The server can also be run via Docker or deployed manually to Azure.

Claude Desktop / Cursor

Paste into your MCP client config file to install this server.

{
    "mcpServers": {
        "business request server": {
            "br-mcp-server": {
                "command": "uv",
                "args": [
                    "venv"
                ]
            }
        }
    }
}

McpServers

{
    "br-mcp-server": {
        "command": "uv",
        "args": [
            "venv"
        ]
    }
}

Business Request Server

Demo PoC MCP server to be used in other PoC

Devs

uv venv
uv pip install -e .

then run it locally

mcp dev server.py

or alternatively

python server.py

Navigate to the URL it showed to test your server.

And then you can test functions such as Templates, and then search_business_requests:

Pass in this:

{
  "query_filters": [
    {
      "name": "BR_SHORT_TITLE",
      "value": "Server",
      "operator": "="
    }
  ]
}

And then you can filter on the results via filter_results:

[
  {
    "column": "RPT_GC_ORG_NAME_EN",
    "operator": "contains",
    "value": "Correctional"
  }
]

Running via Docker

docker build -t mcp-bits:local .
docker run -p 8080:8080 --env-file ./.env --name mcp-bits-container mcp-bits:local

pymssql issues

pymssql on Mac OSX

pymssql has dependency with FreeTDS, as such ensure you install it beforehand brew install freetds.

After which if you have issues with running the code please do the following:

uv pip uninstall pymssql
uv pip install --pre --no-binary :all: pymssql --no-cache --no-build-isolation

Also you can add to uv pyproject.toml

[tool.uv]
no-binary-package = ["pymssql"]

After this all should be working.

NOTE: Known issue with cython==3.1.0 found here

Here is how to get around it for now (please remove this once this issue is fixed):

export CFLAGS="-I$(brew --prefix freetds)/include"
export LDFLAGS="-L$(brew --prefix freetds)/lib"
uv pip install "packaging>=24" "setuptools>=54.0" "setuptools_scm[toml]>=8.0" "wheel>=0.36.2" "Cython==3.0.10" "tomli"
uv pip install --pre --no-binary :all: pymssql --no-cache --no-build-isolation

Deployment

CI/CD

TODO

Manual

This is how you can deploy manually in Azure via the CLI.

az webapp deployment source config-local-git \
  --name <WebAppName> \
  --resource-group <ResourceGroupName>
git remote add azure <GitURLFromPreviousStep>
git push azure main

Documentation

Using this as tutorial on how to build the demo.
FastMCP documentation

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.