Skip to content

codeaihub.in

AI, Finance, Stock, Cloud, and JavaScript: Your Digital Toolbox. Some article content may be AI generated.

MCP Prompts — Questions & Answers

September 14, 2026 By Tech Writer

Q1. What is an MCP Prompt?

Answer:

An MCP Prompt is a reusable prompt template exposed by an MCP server.

It provides structured instructions/messages that an MCP client can retrieve and use in an AI interaction.

Example:

ride_planning_prompt

could provide instructions for planning a cycling ride.

Q2. Is an MCP Prompt the same as the prompt that I type to Claude?

Answer: No.

They are two different things.

User prompt

You might say:

“Can you plan a 100 km ride from Pune?”

That’s your request.

MCP Prompt

The CycleConnect MCP server might provide:

ride_planning_prompt

with instructions such as:

You are a cycling ride planning assistant.

Consider:
- distance
- elevation
- rider experience
- rest stops
- hydration
- safety
- duration

So:

User Prompt
    ↓
"What do I want?"

MCP Prompt
    ↓
"How should this type of task be approached?"

Q3. Do I have to type the MCP Prompt name when talking to Claude?

Answer: No.

The user normally doesn’t need to say:

Use ride_planning_prompt

The user can naturally say:

“Can you plan a 100 km ride from Pune?”

The AI host/client can determine whether an available MCP Prompt is relevant.

Q4. Then how does Claude know that ride_planning_prompt is relevant?

Answer:

The MCP server exposes the prompt and its metadata.

For example:

Name:
ride_planning_prompt

Description:
Plan a cycling ride based on distance,
location and rider experience.

The MCP client can discover available prompts.

Conceptually:

MCP Server
    │
    │ prompts/list
    ▼
MCP Client / AI Host
    │
    ▼
ride_planning_prompt

Claude can then use its understanding of the user’s request to determine whether that prompt is appropriate.

Q5. Does MCP itself understand the user’s sentence and automatically select the Prompt?

Answer: No.

This is an important architect-level distinction.

MCP provides the standardized protocol for discovering and retrieving prompts.

The AI host/client is responsible for understanding the user’s intent and deciding whether to use a particular prompt.

                    User
                      │
                      │
              "Plan a 100 km ride"
                      ▼
                 AI Host
                      │
              Understand intent
                      │
                      ▼
              MCP Client
                      │
                      ▼
              MCP Server
                      │
                      ▼
        ride_planning_prompt

Remember:

MCP provides the mechanism; the AI host provides the intelligence for deciding when to use it.

Q6. What information does an MCP Prompt definition contain?

Answer:

Conceptually, it contains:

MCP Prompt
│
├── Name
├── Title
├── Description
├── Arguments (optional)
└── Prompt messages/content

For example:

Name:
ride_planning_prompt

Description:
Plan a cycling ride.

Arguments:
distance
start_location
experience_level

Messages:
"You are a cycling ride planning assistant..."

Q7. Why do we need the description?

Answer:

The description helps the MCP client/AI host understand what the Prompt is intended for.

For example:

Name:
ride_planning_prompt

Description:
Plan a cycling ride based on distance,
location and rider experience.

The description communicates the purpose of the prompt.

It is therefore important to write clear and meaningful descriptions.

Q8. Does the user need to use exactly the same words as the Prompt description?

Answer: No.

Suppose the description is:

Plan a cycling ride based on distance,
location and rider experience.

The user could say:

“I want to go cycling this Sunday for about 100 km.”

The user doesn’t have to repeat the description word-for-word.

The AI host interprets the user’s intent.

Q9. What happens if the Prompt is too specific?

Suppose we define:

Description:
Plan a ride for exactly 100 kilometers.

Then the user says:

“Plan a 200 km ride.”

The prompt definition isn’t very reusable.

A better design would be:

Description:
Plan a cycling ride based on the requested distance.

with an argument:

distance

This allows:

distance = 100

or:

distance = 200

Q10. What are Prompt arguments?

Answer:

Prompt arguments allow the same Prompt template to be reused with different values.

For CycleConnect:

ride_planning_prompt

could have:

distance
start_location
experience_level

The user says:

“Plan a 100 km ride from Pune for an intermediate cyclist.”

The AI host can conceptually supply:

distance = 100 km
start_location = Pune
experience_level = intermediate

The MCP server can then produce the appropriate prompt messages.

Q11. Why are parameterized MCP Prompts better?

Answer:

Because one Prompt can support many scenarios.

Instead of creating:

ride_50km_prompt
ride_100km_prompt
ride_200km_prompt

we can create:

ride_planning_prompt

with:

distance

Then:

50 km
100 km
200 km

can all use the same Prompt.

Q12. What is the role of the detailed prompt content?

Answer:

The detailed prompt contains the actual instructions/messages that guide the AI interaction.

For example:

You are a cycling ride planning assistant.

Consider:
- distance
- elevation
- rider experience
- rest stops
- hydration
- safety
- duration

The metadata describes the Prompt.

The messages contain the actual instructions.

Think:

Metadata
    ↓
"What is this Prompt?"

Prompt content
    ↓
"What instructions should be provided?"

Q13. Does the MCP Prompt itself perform the ride planning?

Answer: No.

The Prompt provides instructions.

It doesn’t necessarily query databases, calculate routes, or create rides.

For example:

Prompt
   ↓
Instructions for planning

Resource
   ↓
Available ride data

Tool
   ↓
Create/join/cancel a ride

These capabilities can work together.

Q14. How do Tools, Resources and Prompts differ?

Answer:

This is one of the most important MCP concepts.

MCP capabilityPurpose
ToolPerform an action
ResourceProvide/read data
PromptProvide reusable interaction instructions

Our CycleConnect example:

Tool:
create_ride

Resource:
rides://12345

Prompt:
ride_planning_prompt

Easy mnemonic:

Tool = DO
Resource = DATA
Prompt = GUIDE

Q15. Can a Tool also read data?

Answer: Yes.

Don’t memorize:

Tool = write
Resource = read

as an absolute rule.

The better distinction is:

Tool
→ executable capability/action

Resource
→ addressable contextual data

For example, our existing:

read_doc_contents

is a Tool even though it reads data.

Q16. What is an MCP Resource?

Answer:

A Resource is an addressable data source exposed by an MCP server.

For example:

docs://documents/report.pdf

could identify a document.

The client can request the resource and receive its content.

Q17. What is a Resource Template?

Answer:

A Resource Template is a parameterized URI pattern.

Example:

docs://documents/{doc_id}

It can represent:

docs://documents/report.pdf
docs://documents/plan.md
docs://documents/deposition.md

One template can therefore represent many concrete resources.

Q18. How does a Resource differ from a Prompt?

Answer:

A Resource provides data.

A Prompt provides instructions.

For CycleConnect:

Resource:
rides://12345

→ "Here is the information about ride 12345."

Prompt:
ride_planning_prompt

→ "Here is how you should approach planning a ride."

Q19. How can all three work together in CycleConnect?

Answer:

Suppose you say:

“Find me a suitable 100 km ride from Pune and help me join it.”

The architecture could look like:

                         USER
                           │
                           │
                  Natural language
                           │
                           ▼
                        CLAUDE
                           │
             ┌─────────────┼─────────────┐
             │             │             │
             ▼             ▼             ▼
          PROMPT       RESOURCES       TOOLS
             │             │             │
        "How to plan"   "Ride data"   "Join ride"
             │             │             │
             └─────────────┼─────────────┘
                           ▼
                     CycleConnect

For example:

Prompt:
ride_planning_prompt

Resource:
rides://upcoming

Tool:
join_ride

Q20. Does the user have to know MCP exists?

**Answer: Not necessarily.

A well-designed application can hide most of the MCP complexity from the user.

The user simply says:

“Plan a 100 km ride from Pune.”

The application/AI host can handle:

Intent understanding
       ↓
Select appropriate Prompt
       ↓
Read relevant Resources
       ↓
Call required Tools
       ↓
Return result

This is one reason MCP is useful for building AI-powered applications.

Q21. What is the role of MCP Client in this process?

Answer:

The MCP Client is the component that communicates with the MCP server.

Conceptually:

Claude / AI Host
       │
       ▼
   MCP Client
       │
       ▼
   MCP Server

The client can discover:

Tools
Resources
Prompts

and request them from the server.

Q22. What does prompts/list do?

Answer:

It allows the MCP client to discover the prompts exposed by the MCP server.

Conceptually:

MCP Client
     │
     │ prompts/list
     ▼
MCP Server
     │
     ▼
Available prompts

For our project:

ride_planning_prompt

could appear in the result.

Q23. What does the Prompt actually return?

Answer:

A Prompt returns messages.

Our example uses:

messages: [
  {
    role: "user",
    content: {
      type: "text",
      text: "You are a cycling ride planning assistant..."
    }
  }
]

So conceptually:

Prompt
  ↓
Messages
  ↓
AI interaction

Q24. Why would an enterprise want MCP Prompts?

Answer:

Because an organization can centralize reusable AI interaction patterns.

For example:

CycleConnect MCP Server
│
├── ride_planning_prompt
├── safety_analysis_prompt
├── ride_summary_prompt
└── club_recommendation_prompt

Different MCP clients can discover and use these standardized prompts.

This can improve:

  • Consistency
  • Reusability
  • Maintainability
  • Governance
  • Standardization

Q25. What is the most important architect-level statement about MCP Prompts?

Answer:

MCP standardizes how prompts are exposed, discovered and retrieved; the AI host/client determines when a prompt is relevant to the user’s request and how it should be used.

This is more accurate than saying:

“MCP automatically matches the user’s question to a prompt.”

Final Exam Mental Model

When you receive a user request:

"Plan a 100 km ride from Pune."

think:

                 USER
                   │
                   │ Request
                   ▼
                 CLAUDE
                   │
          Understand intent
                   │
       ┌───────────┼───────────┐
       ▼           ▼           ▼
     PROMPT     RESOURCE      TOOL
      GUIDE       DATA         DO
       │           │           │
       │           │           │
       └───────────┼───────────┘
                   ▼
             Final response

The four questions to ask yourself

What does the user want?

→ User Prompt

How should this type of task be approached?

→ MCP Prompt

What information is available?

→ MCP Resource

What action can the AI perform?

→ MCP Tool

That is the core mental model I recommend you use for the rest of the MCP certification course.

Leave a Reply

Your email address will not be published. Required fields are marked *