AiVibeModel Context Protocol (MCP)
The Model Context Protocol (MCP) is an open protocol, introduced by Anthropic in 2024, that standardises how AI applications connect to external tools, data sources and services through MCP servers, instead of requiring a custom integration for each.
MCP uses a client-server architecture. An AI application, the host, runs MCP clients that connect to MCP servers, each exposing capabilities such as tools the model can call, resources it can read and reusable prompts. Messages use JSON-RPC 2.0 over transports such as standard input and output for local servers or HTTP for remote ones. The model sees tool descriptions and decides when to call them, and the host executes the calls.
MCP has been adopted by many AI assistants, development tools and agent frameworks, and servers exist for file systems, databases, code repositories, ticketing systems and business applications. Organisations use it to give assistants controlled access to internal systems through a consistent interface.
An MCP server acts with whatever permissions it is given, so security depends on authentication, least-privilege access, user confirmation for sensitive actions and vetting of third-party servers. Tool descriptions and returned content can carry prompt injection. The specification defines an authorisation approach based on OAuth for remote servers.
Key points
- Open protocol connecting AI applications to tools and data
- Servers expose tools, resources and prompts to clients
- Uses JSON-RPC 2.0 over local or HTTP transports
- Security relies on least privilege, confirmation and vetted servers
Where AiVibe comes in
AiVibe delivers chatbots and virtual assistants with RAG, MCP tools and voice.
Related terms
- Function Calling (Tool Use)AI & Machine Learning
- AI AgentAI & Machine Learning
- Agent2Agent Protocol (A2A)AI & Machine Learning
- Retrieval-Augmented Generation (RAG)AI & Machine Learning
- Prompt InjectionAI & Machine Learning