跟讀練習: Demystifying MCP | Architect Insights - 通過影片學習英語口語
選擇練習模式
每種模式都會開啟全屏練習室。
128 句
如果句子過短或過長，請點選 Edit 進行調整。
1
Hi, I'm Scott Reila, Principal Architect Evangelist at Salesforce.
2
In this episode of Architect Insights, we're going to be exploring Model Context Protocol, or MCP.
3
Often called the USBC for AI, MCP is an open standard revolutionizing how AI agents dynamically connect to external tools and data.
4
Today, we will unpack how it works, examine the architectural trade-offs, and discuss why this new protocol complements your standard APIs.
5
Let's get started.
6
Before we dive in, please note that we may be discussing some forward-looking products and features today.
7
Please make all purchasing decisions based on what is currently available.
8
There are two types of MCP interactions at Salesforce.
9
On the left side, we have Agent Force acting as a client.
10
These are Agent Force agents
11
that need to connect to other external systems to get access to information like order status or inventory availability, or even to execute tasks based on the conversation.
12
On the right side, Salesforce acts as a server.
13
In this scenario, we have external or local agents that need to be able to use Salesforce data via Headless 360.
14
We'll be touching on both sides in our discussion today.
15
But before we go any further, we need to understand exactly what the model context protocol is.
16
MCP is the new open standard for connecting AI to outside tools and data.
17
It was introduced by Anthropic in November of 2024 to solve major integration bottlenecks.
18
Traditionally, if you wanted an agent to interact with a system, you would have to build custom skills or code API or SOC will calls for every single interaction.
19
MCP changes this by providing one unified connection standard, eliminating the need for those custom agent specific builds.
20
It allows AI models to consistently discover and use external features, and because it is open source, it works seamlessly across all AI providers.
21
MCP is often called the USBC for AI.
22
Just like you no longer want to carry around separate micro USB and Lightning cables,
23
MCP means we finally have a single unified configuration protocol for all of our AI agent integrations.
24
Now that we know the what, let's talk about how MCP actually works.
25
Let's start with the component view.
26
Looking at the left side of the screen, we have our hosts.
27
These could be chat interfaces like Claude or ChatGPT, code editors like VS Code or Claude Code or Cursor, or agents like AgentForce.
28
Each of these hosts has its own LLM connection and an embedded MCP client.
29
You can think of the host and the embedded client as sitting together as one unit for now.
30
In the middle, we have the MCP server.
31
The server handles the information discovery and setup, illustrating the different tools, resources, and prompts that are available to the host.
32
There are three types of interactions the MCP server provides.
33
First, we have tools, basically any callable code like APEX, API, whatever.
34
Tools are called dynamically by the agent.
35
For example, if I ask for an order status, the agent will gravitate towards an order status tool to make that call.
36
If there's no appropriate a tool, the LLM will tell me that it can't get the information, or worse, it may actually try to hallucinate the answer.
37
Guardrails are very important here.
38
Next are resources.
39
These represent data or files, and they are fundamentally different from tools because they are not directly callable by the agent.
40
To use the resource, the user must specifically mention it in their question, such as asking for a specific order using a manual pointer like at order 123.
41
Additionally, resources can handle notifications and follow-up information, which only trigger if the underlying system explicitly tells the MCP server that the resource was updated.
42
Lastly, we have prompts.
43
These are essentially just prompt templates.
44
It is critical to note that there is no LLM interaction on the MCP server side.
45
All LLM processing happens over on the host side.
46
Now that we know the components, let's discuss initialization.
47
Anytime you add a new mcp server
48
or start your agent an initialization step occurs exposing an mcp server without security is a bad idea even
49
if the data isn't highly sensitive like a weather app for
50
instance you absolutely must protect your server from denial of service attacks
51
if someone asks an open server for the weather in every
52
single us zip code it could easily crash your mcp server the admin or user
53
if local pre-configures the api key in the headers of the
54
host mcp json config file the host initiates the connection sending a request to the remote mcp server
55
while passing those authorization headers the server then validates the key
56
and if it is a valid api key the server successfully returns the tool list
57
and capabilities for the host to store we'll talk more about storage a little bit later
58
if the key is invalid
59
or missing the server immediately rejects the call with a 401 unauthorized
60
or a 403 forbidden error this step is essential
61
because it allows the mcp server to to identify exactly who is making the call, enabling the server to enforce strict rate limiting or resource throttling.
62
Now let's talk about user authentication, which is what Salesforce uses for Headless 360.
63
This one starts off the same way, where the user or admin configures their host.
64
Looking at step one, the initial connection attempt goes from the client host to the remote MCP server.
65
However, it returns a 401 unauthorized, which prompts the host to fire up a browser and go to the token URL.
66
This takes us to step two where the user logs in through a PKCE or Pixie challenge.
67
Once the user authenticates, Salesforce uses the external app to grant your scopes and determine which MCP servers your user can access.
68
Moving down to step three, the code is exchanged for a bearer and access token.
69
The host connects the user using that bearer token, And as long as the user has the right scopes,
70
the server returns the tool list and capabilities that the user is allowed to access.
71
As you can see, this is a little bit more involved.
72
But this is exactly where we start talking about why it is so important to know who the user is.
73
Every user in Salesforce has different permissions and capabilities.
74
And in this OAuth flow, it ensures that our agent
75
and our chat client can only execute what is specifically available to
76
that user and only access the MCP servers that are enabled.
77
Now that the MCP is initialized, let's talk about the actual tool invocation.
78
Let's say a user asks a question like, what's the humidity level in Chicago?
79
Starting on step one of the diagram, the host app sends the user's prompt alongside a list of available tools stored in the host to the LLM.
80
This is a critical detail depending on how many MCP servers you have connected.
81
The first call can consume a massive amount of your context window
82
because every single associated tool is sent to the LLM with the question in context.
83
The LLM then evaluates the tool descriptions to decide which one to call.
84
Once the tool is selected, like get current weather, we move to step two.
85
The host sends the tool call request, including inferred arguments like the Chicago zip code, to the MCP server.
86
The server executes the action with the external system and then returns the raw data back to the host.
87
Finally, in step three, the host passes the original question alongside that raw tool result back to the LLM.
88
The LLM extracts the specific data, like finding the 76% humidity from a large block of data, and generates the final natural language response for the user.
89
Notice the token burn happening here.
90
We are making multiple LLM calls per interaction, consuming tokens during both the tool selection phase and the final answer generation phase,
91
which is why managing your context window is very important.
92
Now that we have an idea of what MCP call looks like, let's check out a quick demo.
93
I have an MCP server for weather connected right here.
94
When I initialize my connection, look at the list of tools that populate.
95
We have get current weather, get weather conditions, get current air quality, and so on.
96
Notice that there isn't a tool in this list that explicitly mentions humidity levels.
97
This means the LLM actually has to go through each of the tool descriptions to understand
98
which one is the best fit for our question.
99
In this case, it decides get current weather is the best call.
100
Now let's look at the contract happening under the covers.
101
You can see the tool is looking for a location parameter, like Chicago.
102
When we run this, it goes out to the external system and returns an answer, but it doesn't just return a clean 53%.
103
Look at this large blob of data right here.
104
This is the entire result set that gets passed back to the LLM.
105
If I scan through it, I can see the humidity level is 53%, buried right here.
106
In order for the LLM to give that simple answer, the LLM has to process this entire raw result set alongside the user's original question and context.
107
This is where more token burn happens.
108
You are consuming a large amount of tokens based on the
109
sizable tool result just to extract one specific data point and generate a final natural language answer.
110
For an agent interaction, this is necessary to give the exact answer.
111
Is the same level of processing and token burn needed for a system-to-system integration?
112
Based on this analysis, we need to understand each use case and determine if MCP or API should be used.
113
Let's take a look at a side-by-side comparison of why both exist.
114
Looking at the left side, APIs are built for integration.
115
They rely on programmatic code and deterministic contracts.
116
Because there is no LLM interpreting the request or response, there is zero token cost and the data structure is strict and static.
117
Now looking at the right side, MCP is built for interaction.
118
It relies on the probabilistic reasoning of an LLM, which as we saw in the demo, consumes tokens to select a tool and generate the final answer.
119
However, MCP has dynamic discovery.
120
If you add a new tool to your server, the agent knows it exists after the next initialization
121
and how to use it without you having to build out custom agent-specific integrations.
122
For AgentForce, this is slightly different because the admin has to enable each of the tools for security reasons.
123
For system-to-system integrations, APIs remain as the go-to standard due to their deterministic nature and strict processing on the client side.
124
However, if natural language processing or additional reasoning is needed, like in an agentic interaction, then MCP would be desirable.
125
Ultimately, MCP is a powerful standard that complements your APIs.
126
It does not replace them.
127
By understanding the strengths and limitations of both, you can build smarter agentic interactions while maintaining the reliability of your system to system integrations.
128
Thanks for watching and see you next time.
FOR YOU
推薦影片
C11
497 句
Technical deep dive on Managed Instance for App Service | BRK102
Daily ConversationWorking Adults
C11
392 句
Ultimate Claude Guide: How to Use Claude AI for Beginners in 2026
Daily ConversationWorking Adults
C11
364 句
Running AI Agents in Secure Sandboxes with E2B & Docker MCP | Docker’s AI Guide to the Galaxy
Daily ConversationUniversity StudentsWorking Adults
C11
114 句
Harnessing generative AI in architecture, engineering & construction
Daily ConversationUniversity StudentsWorking Adults
C12
82 句
CAP Theorem Simplified
Daily ConversationUniversity StudentsWorking Adults
C11
346 句
Trust the Process: How to Design Teams That Actually Work
Daily ConversationUniversity StudentsWorking Adults
C11
199 句
十年後的電腦會長什麼樣？黃仁勳：就像今天的手機，最少拿來打電話｜中英字幕
Daily ConversationUniversity StudentsWorking Adults
C11
143 句
MCP vs API: The protocol every developer needs to know
Business EnglishUniversity StudentsWorking Adults
C11
124 句
Agentic AI Frameworks Explained: Workflows, Multi-Agent, & Production
Daily ConversationUniversity StudentsWorking Adults
C11
853 句
Why Coding Agents Keep Making Your Codebase Worse
Daily ConversationWorking Adults
C15
107 句
What is NLP? Neuro Linguistic Programming
Daily ConversationWorking Adults
C1101 句
BGP Overview
Daily ConversationUniversity StudentsWorking Adults
真實場景：AI技術對話中的專業表達
這段影片是一場關於AI領域Model Context Protocol（MCP）的專業對話，涉及技術原理、元件介紹等內容。類似的場景常見於科技講座、行業研討會或工作中的技術交流。通過學習這樣的對話，你能掌握如何清晰表達複雜概念，提升在專業場景中的英語溝通能力，這對雅思口語練習或職場英語都很有幫助。
實用短語與搭配
- revolutionizing how...：徹底改變……的方式。例：MCP is revolutionizing how AI agents connect to external tools.（MCP正在徹底改變AI代理連線外部工具的方式。）
- unified connection standard：統一連線標準。例：It provides a unified connection standard for AI integrations.（它為AI整合提供了統一的連線標準。）
- handle information discovery：處理資訊發現。例：The server handles information discovery and setup.（伺服器負責資訊發現和設定。）
- trigger notifications：觸發通知。例：Resources can trigger notifications when updated.（資源更新時可觸發通知。）
- exposing... without security：暴露……而沒有安全措施。例：Exposing an MCP server without security is a bad idea.（暴露MCP伺服器而沒有安全措施是個壞主意。）
你的影子跟讀挑戰
現在就行動起來！開啟影片，找到“Now that we know the what, let's talk about how MCP actually works.”這段，開始英語影子跟讀。先聽一遍，注意說話人的語調、重音和停頓；再跟著讀，儘量同步。重複3次後，用手機錄下自己的發音，對比原音，看看哪裡需要改進。這個方法能快速提高英語發音，還能鍛鍊聽力和口語流暢度。記住，shadow speech的關鍵是“模仿”，越像越好！每天花5分鐘，堅持一週，你會發現自己的口語進步看得見。
通過看YouTube學英語，結合影子跟讀，你不僅能學到專業詞彙，還能掌握地道的表達節奏。趕緊試試吧，下一個口語達人就是你！
什麼是跟讀法？
跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧，最初開發用於專業口譯員的培訓，並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大：您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同，跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性，語調，節奏，連讀，聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。