Model Context Protocol (MCP)

An open protocol (developed by Anthropic) that standardizes how AI applications connect to external data sources and tools. MCP defines a client-server architecture where AI models (clients) can discover and invoke capabilities provided by MCP servers (search a database, call an API, read a file system, validate a document) through a consistent interface.

Think of MCP as a USB-C port for AI tools. It doesn't matter who made the tool or what it does internally, as long as it speaks the protocol. Before MCP, every tool connected to every model in its own bespoke way. It was the software equivalent of a drawer full of incompatible charger cables.

Why it matters for writers: MCP is how AI agents gain capabilities beyond text generation. LlmsTxtKit ships as an MCP server, which means any MCP-compatible AI agent can use it to fetch and validate llms.txt files as part of a larger workflow. For technical writers, MCP is also interesting because documenting MCP server capabilities (tool descriptions, parameter schemas, expected behaviors) is a documentation challenge that barely anyone has tackled yet. Early mover advantage for tech writers willing to learn the territory.

Related terms: MCP Server · AI Agent · Tool Use · Schema