What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard for connecting AI applications to tools and data, and a new attack surface for SOCs to understand.
Definition
Model Context Protocol (MCP)
Also called MCP, MCP server
The Model Context Protocol (MCP) is an open standard that defines how AI applications connect to external tools, data sources, and workflows, so a model can read context and take actions through a common interface.
How it works
The MCP documentation describes it as an open-source standard for connecting AI applications to external systems, comparing it to a USB-C port: one standard way for applications like Claude or ChatGPT to reach data sources, tools, and workflows. The specification takes inspiration from the Language Server Protocol and uses JSON-RPC 2.0 messages between three roles: hosts (the LLM applications that start connections), clients (connectors inside the host), and servers (services that provide context and capabilities).
Servers can offer resources (context and data), prompts (templated messages and workflows), and tools (functions the model can execute). Clients can offer servers features such as sampling, which lets a server request model completions. A security team might run a server that exposes "search logs" and "get alert details" tools, so any MCP-capable assistant can use them without a custom integration.
The specification acknowledges that MCP enables arbitrary data access and code execution paths. It requires explicit user consent for data access and tool use, and its security best practices document covers confused deputy attacks through OAuth proxy servers, token passthrough (accepting tokens not issued to the server), server-side request forgery, session hijacking, and compromise of local MCP servers.
Example
A detection engineer working in Claude Code asks for a detection of OAuth consent grants to unverified apps. Through an MCP server, the agent queries recent identity logs to see the field names, writes the rule, backtests it, and opens it for review. Cotool's MCP server supports this workflow, letting Claude Code, Codex, Cursor, or any agent harness write, test, and deploy detections from the editor, and its SIEM exposes the same logs to a team's own agents over MCP.
Why it matters for a SOC
MCP cuts both ways for a SOC. It makes security data and tools usable by any AI agent, which speeds up investigation and detection work. It also creates new things to monitor: employees connecting assistants to corporate data through unvetted servers (shadow AI), servers holding powerful tokens, and agents that may follow instructions hidden in tool output (prompt injection). Inventorying MCP servers, scoping their permissions, and logging tool calls is quickly becoming part of SOC coverage.
Related terms
- AI agentAn AI agent is a system in which a large language model dynamically directs its own process and tool use to complete a task, deciding which steps to take rather than following a fixed, predefined script.
- Prompt injectionPrompt injection is an attack in which text supplied to a large language model, directly by a user or hidden in content the model reads, alters its behavior in ways its developers didn't intend, such as ignoring instructions, leaking data, or misusing tools.
- Shadow AIShadow AI is the use of AI tools, models, or services by employees without the approval or oversight of the organization's IT or security teams.
- Detection engineeringDetection engineering is the practice of designing, building, testing, deploying, and maintaining the rules and analytics that turn security telemetry into alerts, treated with the rigor of software engineering.
Go deeper
- What is an AI SOC?An AI SOC uses AI agents to triage, investigate, and close or escalate security alerts with evidence. How it works, how it differs from SOAR, and the risks.
- Security MCP servers: what the official ones can doWho publishes each security MCP server, where it runs, how it signs in, and what it can read or change: Splunk, CrowdStrike, Sentinel, Google SecOps, and more.
- How to connect security tools to Claude CodeAdd Splunk, SentinelOne, and other security MCP servers to Claude Code: the commands, scopes, read-only setup, and where these connections fall short.
FAQ
- What are the main security risks of MCP?
- The MCP specification's security guidance covers confused deputy attacks through OAuth proxy servers, token passthrough, server-side request forgery, session hijacking, and compromised local MCP servers. Prompt injection through tool outputs and over-permissioned tools are the other common concerns, because an agent may act on whatever text a tool returns.
Sources
- What is the Model Context Protocol (MCP)? - Model Context Protocol · accessed November 16, 2026
- Specification (2025-06-18) - Model Context Protocol · accessed November 16, 2026
- Security Best Practices - Model Context Protocol · accessed November 16, 2026
- Detection Engineering Platform | Cotool · accessed November 16, 2026
- AI-Native SIEM | Cotool · accessed November 16, 2026
