Model Context Protocol (MCP)
An open standard that lets AI applications connect to external tools, data sources, and services through a single common interface, instead of every tool and every model needing custom integrations.
The Model Context Protocol (MCP) is an open standard, originally introduced by Anthropic, for connecting AI assistants and agents to outside systems. It defines a client–server protocol: an MCP client (the AI app) discovers and calls capabilities exposed by MCP servers (tools, resources, and prompts).
Before MCP, connecting a model to GitHub, Slack, a database, or a local filesystem required bespoke glue code for each combination. MCP turns those integrations into reusable servers that any compliant client can use. A single filesystem server, for example, can serve every compatible assistant.
Core concepts:
- Tools: callable functions the model can invoke, with typed inputs.
- Resources: read-only data the client can attach to context.
- Prompts: reusable, parameterised instructions.
MCP has been adopted across developer tools, IDEs, and agent frameworks, and is often described as “USB-C for AI tools.” Security is an active area of work: because servers can expose powerful capabilities, prompt injection and permission scoping are key concerns in production deployments.