MCP connectors

MCP connectors are standardized interfaces that link AI models to local data sources and third-party software tools.

The Model Context Protocol (MCP) replaces fragmented integrations with a universal standard for AI connectivity. These connectors allow Large Language Models (LLMs) to securely query local databases like PostgreSQL, interact with GitHub repositories, or pull live data from Slack and Google Drive. By decoupling the data source from the model interface, developers use a single protocol to swap between providers like Anthropic, OpenAI, or local instances without rewriting integration logic. It is a production-ready framework (built on JSON-RPC 2.0) that gives agents direct, governed access to the specific files and tools they need to execute complex workflows.

https://modelcontextprotocol.io