What is an MCP Prompt?

2 min read Updated

An MCP prompt is a reusable, parameterised prompt template exposed by an MCP server that provides standardised workflows and interaction patterns for AI agents — like templates for common tasks.

WHY IT MATTERS

MCP prompts are the third capability type alongside tools and resources. While tools enable actions and resources provide data, prompts provide templates for common agent interactions — standardising how agents approach recurring tasks.

A prompt might define a workflow like 'deploy to staging' with parameters for the branch, environment, and approval status. The agent fills in the parameters and gets a well-structured prompt optimised for that specific task.

Prompts are particularly useful for complex workflows that require specific sequencing or domain knowledge. Rather than hoping the agent figures out the right approach, prompts encode best practices into reusable templates.

HOW POLICYLAYER USES THIS

Intercept can enforce governance over prompt template usage. YAML policies can restrict which prompt templates an agent is permitted to invoke, and constrain the parameters passed to them. This prevents agents from using workflow templates that exceed their authorised scope — for example, allowing a 'deploy to staging' prompt but denying 'deploy to production'.

FREQUENTLY ASKED QUESTIONS

How are MCP prompts different from system prompts?
System prompts configure the agent's overall behaviour. MCP prompts are task-specific templates served dynamically by MCP servers. Think of system prompts as the agent's personality and MCP prompts as standardised workflows it can invoke.
When should I use MCP prompts vs tools?
Use tools when the agent needs to take an action (write a file, run a query). Use prompts when you want to guide the agent through a specific reasoning workflow or standardise how it approaches a type of task.
Can Intercept enforce policies on prompt invocations?
Yes. Intercept treats prompt invocations as policy-evaluable events just like tool calls. You can allow or deny specific prompts and constrain their parameters in YAML policies.

FURTHER READING

Enforce policies on every tool call

Intercept is the open-source MCP proxy that enforces YAML policies on AI agent tool calls. No code changes needed.

npx -y @policylayer/intercept
github.com/policylayer/intercept →
// GET IN TOUCH

Have a question or want to learn more? Send us a message.

Message sent.

We'll get back to you soon.