Model Context Protocol

The Model Context Protocol (MCP) is an open standard, introduced by Anthropic, that defines how AI applications connect to external tools and data sources. Before it, every integration between an assistant and a system such as a database, issue tracker, or file store was custom work repeated separately for each product. MCP replaces that with a common interface: a server exposes tools, resources, and prompts over a defined protocol, and any compliant client can discover and call them. A server written once for an internal system therefore works across different AI clients. Anthropic published the specification and reference SDKs openly, support has spread to other vendors and to editors such as Claude Code, Cursor, and VS Code, and community servers cover GitHub, Postgres, Slack, Playwright, and many other systems. In practice a developer registers a server in the client's configuration and the model can then call its tools during a conversation. MCP standardizes plumbing, not judgment, so the security burden stays with the operator. A server granted broad write access can act destructively when a model misreads intent, and untrusted tool output can carry prompt injection, which is why scoped credentials, approval steps, and audit logs remain necessary.