Open iT / ServiceNow MCP Pilot
This MCP server provides a read-only tool to find products without entitlement records using deterministic mock data, supporting publisher filtering and result limits. It is a pilot scaffold for Open iT and ServiceNow SAM Pro integration.
README
Open iT / ServiceNow MCP Pilot
This repository contains a deliberately small remote MCP server scaffold for an Open iT / ServiceNow SAM Pro pilot. It exposes one read-only tool over Streamable HTTP and uses deterministic mock inventory data; it does not call Open iT or ServiceNow APIs.
What it exposes
- MCP endpoint:
/api/mcp - Public health endpoint:
/api/health - Tool:
find_products_without_entitlements
The MCP endpoint follows the Microsoft Learn MCP pattern: clients use Streamable HTTP, while a normal browser GET receives HTTP 405 Method Not Allowed and a short explanation. The endpoint is stateless and returns JSON MCP responses, which keeps it suitable for horizontally scaled or serverless hosting later. api/index.py exposes the same ASGI application as a future Vercel entry point; no deployment is included.
Run locally
Python 3.11 or newer is required.
python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -e ".[dev]"
$env:OPENIT_MCP_API_KEY = "local-dev-key"
$env:MCP_BIND_HOST = "127.0.0.1"
$env:PORT = "8000"
$env:MCP_BASE_URL = "http://127.0.0.1:8000"
openit-mcp
For bash/zsh, activate with source .venv/bin/activate and configure the key with:
export OPENIT_MCP_API_KEY="local-dev-key"
export MCP_BIND_HOST="127.0.0.1"
export PORT="8000"
export MCP_BASE_URL="http://127.0.0.1:8000"
MCP_BIND_HOST and PORT are used only by the local openit-mcp Uvicorn launcher. MCP_BASE_URL is used only by external clients such as examples/invoke.py; the server itself routes internally with relative paths and never calls itself through an absolute URL.
The server intentionally returns 503 from MCP POST and DELETE requests when OPENIT_MCP_API_KEY is unset. Clients should send the key as a bearer token:
Authorization: Bearer local-dev-key
For ServiceNow compatibility, the server also accepts the raw key in Authorization as well as X-API-Key: local-dev-key. The health endpoint does not require authentication.
Check health:
Invoke-RestMethod "$env:MCP_BASE_URL/api/health"
In a second terminal with the same virtual environment and API key, run the included MCP client example:
python examples/invoke.py
You can also connect an MCP Inspector or other Streamable HTTP client to $MCP_BASE_URL/api/mcp and configure the Authorization header shown above.
Opening the MCP URL directly in a browser returns:
HTTP 405 Method Not Allowed
This is an MCP server endpoint and cannot be accessed directly via a browser. Please use a streamable HTTP MCP client.
Tool contract
find_products_without_entitlements accepts:
| Input | Type | Required | Behavior |
|---|---|---|---|
publisher |
string | No | Exact, case-insensitive publisher filter |
limit |
integer | No | Defaults to 20; allowed range is 1 to 100 |
It returns structured JSON in this shape:
{
"count": 1,
"items": [
{
"product_name": "Open iT Analyzer Pro",
"publisher": "Open iT",
"reason": "No entitlement record found"
}
],
"explanation": "These products currently do not have matching entitlement records in the mock dataset.",
"generated_at": "2026-08-05T12:00:00Z"
}
Results are sorted by publisher and product name before limit is applied.
Tests
pytest
The tests verify the published tool list, structured output, publisher filtering, deterministic limiting, browser 405 behavior, real HTTP MCP initialization/list/call requests, the public health endpoint, and fail-closed API key behavior.
Future hosted configuration
Loopback addresses are local-only: inside Vercel they refer to the function instance, not the public deployment or another service. The network-base address ending in .0 at the start of the IPv4 loopback block is not an application host and must not be used. Hosted URLs must come from deployment environment variables, while same-app routes should remain relative.
Vercel imports the ASGI app from api/index.py; it does not execute the local Uvicorn launcher, so MCP_BIND_HOST and PORT are not deployment assumptions. Configure these server environment variables before hosting:
OPENIT_MCP_API_KEY=<production-secret>
# Recommended when using a stable custom domain:
OPENIT_MCP_ALLOWED_HOSTS=openit-mcp.example.com,openit-mcp.example.com:*
The server automatically adds VERCEL_URL, VERCEL_BRANCH_URL, and VERCEL_PROJECT_PRODUCTION_URL to its MCP host allowlist when Vercel exposes those system variables. An explicit OPENIT_MCP_ALLOWED_HOSTS remains recommended for a stable custom domain.
Set MCP_BASE_URL=https://openit-mcp.example.com only if a hosted client or job runs examples/invoke.py; the MCP server routes /api/mcp and /api/health without it. Keep the API key in Vercel's secret environment configuration and use HTTPS. Environment changes apply only to new deployments, so redeploy after changing them.
Connect from ServiceNow AI Agent Studio
Use the following values after deploying to Vercel:
| Field | Value |
|---|---|
| Name | Open iT SAM Pilot |
| Authentication type | API Key |
| MCP server URL | https://<your-stable-vercel-host>/api/mcp |
| API key | The raw value of OPENIT_MCP_API_KEY |
Do not add a Bearer prefix in the ServiceNow API Key field. The server accepts ServiceNow's raw Authorization value. A browser request to the URL still returns the intentional 405; tool discovery uses authenticated MCP POST requests.
Before ServiceNow integration
This pilot still needs a real Open iT data adapter, agreement on the product/entitlement matching rules and response fields, ServiceNow remote MCP connection configuration, production authentication (preferably OAuth 2.1 rather than a shared key), HTTPS hosting, logging/monitoring, secret rotation, and end-to-end validation from AI Agent Studio.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.