Open standard for connecting AI models to external tools and data sources through a shared interface. Supported by multiple AI vendors, replacing one-off integrations with a portable contract.
Definition: Open standard for connecting AI models to external tools and data sources through a shared interface. Supported by multiple AI vendors, replacing one-off integrations with a portable contract.
Before MCP, every connection between an LLM and an outside system was bespoke: each transcription service, each analytics dashboard, each spreadsheet had its own integration code. MCP defines a common contract that both the model side and the tool side implement, so the same tool definition works regardless of which model is calling it.
For research, the practical wins are portability and reproducibility. A workflow built against MCP can swap the underlying LLM without rewriting the tool integrations. Tool calls are logged as structured records, which makes it possible to reconstruct exactly what data the model saw and what actions it took.
MCP is open and vendor-neutral. Adoption is growing across major AI vendors as of 2026. For workflows you expect to maintain for more than a quarter, the protocol-first path beats locked-in integrations.
This term is referenced in the following articles: