MCP server
A Model Context Protocol endpoint or process that exposes tools, resources or prompts an agent can call.
- Published
- Sep 27, 2026
- Checked
- Sep 27, 2026
How far we checked this page
Signed by Ravalika Korthiwada, Sep 27, 2026. Passing one check says nothing about the next.
- FormatNot checkedThe file format was not checked for this page.
- InstallsNot checkedInstallation was not checked for this page.
- LoadsNot checkedLive loading was not checked for this page.
- HelpsNot checkedA with/without task comparison was not run.
An MCP server is a process or endpoint that implements the Model Context Protocol and exposes capabilities (tools, resources or prompts) to a connected client. The MCP tools specification, revision 2026-07-28, defines discoverable tools with names and input schemas: a client lists them with tools/list and calls them with tools/call. An MCP server isn’t synonymous with “a tool” alone; the broader MCP specification also covers resources and prompts.
For example, a read-only issue-search server could expose a tool that lists open issues from a tracker, letting an agent skill instruct the agent to fetch and cite live issues rather than only summarizing a file already in the workspace. Claude Code reports MCP connection state with claude mcp list, claude mcp get <name> or /mcp; an “Added” message there confirms configuration was written, not that the server actually connected. Claude Code MCP docs.
We haven’t installed or connected an MCP server ourselves as of this writing, so we can’t speak to connection reliability, context cost, or security behavior from our own testing. See MCP’s own security guidance for documented risks around authorization and local servers.
Related: Agent skill vs MCP server, Agent skill.