A3.5 · LESSON 5 OF 6

Resources and prompts

Explain MCP roles and lifecycle and trace a tool-discovery and tool-call exchange under a pinned protocol version.

THE IDEA

Resources and prompts

Tools are only one of the capabilities a server can offer. Resources expose contextual data (a file, a record) that the host can read and place in context, via resources/list and resources/read. Prompts expose reusable interaction templates with arguments, via prompts/list and prompts/get. They are different from tools: a resource read returns data and changes nothing; a prompt produces messages to send; a tool may act on the world.

purpose: contextual data the host can read and offer as context methods: resources/list, resources/read example: file:///policy/refunds.txt → "Refunds are paid within 30 days." effect: read-only data; reading changes nothing

Reading a resource returns data; it is not the same as executing an arbitrary shell command. Each capability has its own methods and its own risks.

Compare the three capabilities.

Predict first. Is reading a resource identical to executing an arbitrary shell command?

Chapter context

MCP defines a protocol for applications to connect models with external capabilities. It standardizes exchanges; it does not grant a server universal authority or make every discovered tool appropriate for every user request.

TRY IT YOURSELF

Part 1. Do it hands-on

Classify capabilities by hand, then write resource and prompt handlers.

Reading a resource returns data; it is not the same as executing an arbitrary shell command. Each capability has its own methods and its own risks.

policy document to read as context action that sends an email reusable summarise template resource read = shell command? (yes/no)

To do: Classify each as tool, resource or prompt. The text of a policy document the model should read as context
To do: An action that sends an email
To do: A reusable template "Summarise {topic} in {n} bullet points" with arguments
To do: Is reading a resource identical to executing an arbitrary shell command? Type yes or no

Part 2. Do it in code

The envelope helpers (check_envelope, error_response) are PROVIDED as before. The demo server has RESOURCES = {uri: text} and PROMPTS = {name: {"arguments": [...], "template": "... {arg} ..."}}.
resources_list(req): result resources = [{"uri", "name"}] for every resource (name = the uri's last path segment), in the dict's order.
resources_read(req): params.uri → result contents = [{"uri", "text"}]. An unknown uri is a protocol error -32602 "Unknown resource: <uri>". Reading changes nothing.
prompts_list(req): result prompts = [{"name", "arguments"}] for every prompt.
prompts_get(req): params.name and params.arguments (a dict). Unknown prompt → -32602 "Unknown prompt: <name>"; a missing required argument → -32602 "Missing argument: <arg>". Otherwise result messages = [{"role": "user", "content": {"type": "text", "text": the template with each {arg} replaced}}].
handle(text): envelope check, notifications → None, then dispatch resources/list, resources/read, prompts/list, prompts/get; anything else → -32601.

import json

VERSION = "2026-07-28"
VKEY = "io.modelcontextprotocol/protocolVersion"
CKEY = "io.modelcontextprotocol/clientCapabilities"
RESOURCES = {"file:///policy/refunds.txt": "Refunds are paid within 30 days.", "file:///policy/shipping.txt": "Shipping takes 3 to 5 days."}
PROMPTS = {"summarize": {"arguments": ["topic", "n"], "template": "Summarise {topic} in {n} bullet points."}}

def error_response(id, code, message, data=None):
    e = {"code": code, "message": message}
    if data is not None:
        e["data"] = data
    return {"jsonrpc": "2.0", "id": id, "error": e}

def check_envelope(text):
    try:
        req = json.loads(text)
    except Exception:
        return error_response(None, -32700, "Parse error")
    if not isinstance(req, dict) or req.get("jsonrpc") != "2.0" or not isinstance(req.get("method"), str):
        return error_response(req.get("id") if isinstance(req, dict) else None, -32600, "Invalid request")
    if "id" not in req:
        return req
    meta = (req.get("params") or {}).get("_meta") or {}
    if VKEY not in meta or not isinstance(meta.get(CKEY), dict):
        return error_response(req["id"], -32602, "Invalid params: missing required _meta field(s)")
    if meta[VKEY] != VERSION:
        return error_response(req["id"], -32022, "Unsupported protocol version", {"supported": [VERSION], "requested": meta[VKEY]})
    return req

def resources_list(req):
    pass

def resources_read(req):
    pass

def prompts_list(req):
    pass

def prompts_get(req):
    pass

def handle(text):
    pass

Ctrl/⌘ + Enter runs it

Stuck, or something not working? You can come back to it any time.

CHECK YOUR UNDERSTANDING

Is reading a resource identical to executing an arbitrary shell command?

ANoBYes

Chapter reference: math, worked example & code

This section belongs to the whole chapter, so it is the same on every lesson in it and may cover a different lesson's topic. The code for the experiment you just ran is inside the experiment itself.

Math and notation

The MCP specification revision examined here is 2026-07-28. Request envelopes use JSON-RPC 2.0, an ID for response correlation, a method, and method-specific parameters that include _meta with the required io.modelcontextprotocol/protocolVersion and io.modelcontextprotocol/clientCapabilities. A successful result carries resultType, here "complete". Do not treat this date as a claim about every installed client or SDK.

Worked example

A tools/call request with id 7 names add, supplies a=2,b=3 and carries the protocol version and client capabilities in _meta. The server runs the permitted operation and answers with the same id and a content array holding the text 5. The offline example below checks envelope fields and arithmetic; a production server must also implement discovery, schema validation, errors and transport behavior.

Reference code

The following example uses the Python standard library and does not call a remote service.

meta={'io.modelcontextprotocol/protocolVersion':'2026-07-28','io.modelcontextprotocol/clientCapabilities':{}}
request={'jsonrpc':'2.0','id':7,'method':'tools/call','params':{'name':'add','arguments':{'a':2,'b':3},'_meta':meta}}
args=request['params']['arguments']
total=args['a']+args['b']
response={'jsonrpc':'2.0','id':request['id'],'result':{'resultType':'complete','content':[{'type':'text','text':str(total)}],'isError':False}}
assert request['method'] == 'tools/call'
assert request['params']['_meta']['io.modelcontextprotocol/protocolVersion'] == '2026-07-28'
assert response['id'] == 7
assert response['result']['content'][0]['text'] == '5'
assert response['result']['isError'] is False

Python examples are provided for study and local execution; this page does not execute Python.

Common mistakes

MCP is not a language model or an automatic permission system. Resource, prompt and tool capabilities differ. Do not assume an initialize handshake: that belongs to revisions before 2026-07-28. Protocol versions and SDK releases must be matched when turning this draft into runnable setup instructions.