Agent concepts
Model Context Protocol (MCP)
The Model Context Protocol is an open standard that lets AI agents connect to external tools and data sources through a common interface.
Also called: MCP
Definition
What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard for connecting AI applications to external capabilities. An MCP server exposes tools, resources, and prompts; an MCP client, usually an AI agent or assistant, discovers them and calls them over a defined wire format. It plays a role for agents that USB played for hardware: one connector instead of a bespoke integration per tool.
MCP is powerful because it makes an agent's reach pluggable. It is a security concern for the same reason. Any MCP server an agent trusts can describe tools whose names and descriptions the model reads and acts on, which opens the door to tool poisoning, shadowing, and silent changes to a server's behavior after it has been approved.
FAQ
Common questions.
What does an MCP server do?
An MCP server exposes tools, resources, and prompts that an AI agent can discover and call. Examples include servers for a filesystem, a database, GitHub, or a search API.
Is MCP secure by default?
No. MCP defines the connection, not the trust. An agent must still verify which servers it connects to, inspect the tools they declare, and govern the calls it makes, because a malicious or compromised server can influence the agent through the tool definitions it advertises.
Related terms
Keep reading.
Govern the actions, not just the vocabulary
AxioRank scores every tool call your agents make for leaked secrets, PII, destructive operations, and prompt injection, checks it against your policy, and proves it in a tamper-evident audit log. Start free, no card.