MCP server
An MCP server is a program that exposes a system's data and actions to AI assistants through the Model Context Protocol. For test management, it lets an assistant such as Claude or Cursor search test cases, create runs or record results directly, acting with the permissions of the token it was configured with.
Also called: Model Context Protocol server
Where MCP comes from
The Model Context Protocol is an open standard that Anthropic introduced in November 2024 for connecting AI assistants to external tools and data. A server describes what it offers as a set of tools, each with a name and typed inputs, and any compatible client, such as Claude Code, Claude Desktop, Cursor or VS Code, can call them. The protocol replaces one-off integrations with one interface per system.
What it changes in testing work
Without MCP, an assistant only knows what someone pastes into the chat: a ticket, a few test cases, an error log. With an MCP server for a test management system, it can look things up itself. Typical requests are drafting test cases from a ticket into the right suite, finding duplicates, creating a run for a release, or summarising which cases failed and why. The assistant works on the real records, so its output lands where the team already works.
Permissions come first
An MCP server usually acts with the identity and rights of the token it was given. A read-only role stays read-only; a token with write access can change or delete test cases. Decide which role the assistant gets, keep tokens out of version control, and prefer a system that logs every change under a named user. The trade-offs of giving agents write access are covered in giving AI agents write access to your TMS, and a setup walkthrough is in connecting Claude to your TMS via MCP.