t5gmcp

FastMCP-based server that exposes tools to query a T5G dashboard for cards, cases, bugs, escalations, issues, and enriched case data.

Prerequisites

• Python 3.9+
• pip

Install

pip install -r requirements.txt

Configure

Set the base URL for the dashboard API used by the server.

export DASHBOARD_API="https://your-dashboard.example.com/api"

Run the server

• Using the helper script:

./run.sh

• Or directly with fastmcp:

fastmcp run server.py:mcp --transport http --port 8000

The server will listen at http://localhost:8000/mcp by default.

Chatbot

A minimal chatbot is provided that can optionally use an LLM (OpenAI) with tool calling to invoke the MCP tools. Without an LLM, it falls back to keyword routing.

# optional if using LLM-backed chat
export OPENAI_API_KEY=sk-...
export OPENAI_MODEL=gpt-4o-mini

# ensure server is running first (see above)
python chatbot.py --mcp-url http://localhost:8000/mcp

# to run without an LLM and rely on keyword routing
python chatbot.py --no-llm

CLI client

This repo includes a simple client to invoke server tools over HTTP.

python client.py -d cards
python client.py -d cases
python client.py -d bugs
python client.py -d details
python client.py -d escalations
python client.py -d issues
python client.py -d full_case_data

Exposed tools

Server tools are defined in server.py and can be called via MCP or the provided CLI.

• get_cards: Returns all JIRA cards as a dict.
• get_cases: Returns all customer cases as a dict.
• get_bugs: Returns JIRA bugs keyed by case number.
• get_details: Returns extended case details keyed by case number.
• get_escalations: Returns a list of escalated cases.
• get_issues: Returns issues keyed by case number.
• get_full_case_data: All the info from get_cards, but keyed on cases rather than JIRA cards