Linux Foundation / CNCF MCPA: Model Context Protocol Associate (MCPA)

Last checked November 10, 2026 against the official Linux Foundation / CNCF exam page.

1. Question 1Interactions & Execution

   A tool runs, but the external API it calls fails. Which response follows the split between protocol errors and tool execution errors in MCP?

   1. AClose the transport connection so that the client learns the tool has failed
   2. BReturn a JSON-RPC protocol error for every failed API call, so the host stops the conversation and reports to the operator
   3. CReturn a normal tool result flagged with isError and a message describing the failure, so the model can read it and respond
   4. DReturn an empty content list with no error flag, so that the failure does not appear in the model's context
   Show answer and explanationHide answer

   Correct answer: C

   MCP reports tool execution failures inside the tool result with isError set, so the model can see them and possibly retry, while protocol errors cover problems such as malformed requests. Hiding the failure or dropping the connection gives the client and model no usable information.

2. Question 2Security & Governance

   A host is about to connect to a third-party MCP server that the organization has not reviewed. Which principle should guide the decision?

   1. ATrust the server because it implements the standard protocol correctly, so its tools can be used without review
   2. BTreat the server and anything it returns as untrusted until reviewed, and share only the data the task needs
   3. CGive the server full read and write access to the user's home directory so that it can work reliably in every possible workflow
   4. DAssume the model will filter out any malicious content, so the server's outputs need no further checks
   Show answer and explanationHide answer

   Correct answer: B

   Protocol compliance says nothing about whether a server is safe, so unreviewed servers and their outputs should be treated as untrusted and given minimal data. Broad home-directory access and reliance on the model to filter content both enlarge the risk.

3. Question 3Use Cases & Ecosystem

   A business analyst wants an assistant to answer questions about sales figures through an MCP server. Who is best placed to decide which fields the server exposes and which credentials it uses?

   1. AThe end user typing the questions, who can change the server's access list at any time from the chat window
   2. BThe team that owns the sales data and the server, working within the organization's access policies
   3. CThe model provider, which configures every exposed field in its hosted infrastructure automatically
   4. DThe network administrator who manages firewall rules between the host and the internet
   Show answer and explanationHide answer

   Correct answer: B

   Data owners and platform teams decide what a server exposes and which credentials it uses, under the organization's policies. End users, model vendors, and network staff do not own those exposure decisions.

4. Question 4MCP Fundamentals

   A startup wants its internal ticketing system, document store, and code search to be usable by several different AI assistants without writing a separate integration for each assistant. Which approach best matches the purpose of the Model Context Protocol?

   1. ABuild a separate proprietary plugin for each assistant vendor so that every product gets its own bespoke integration code
   2. BFine-tune one model on the ticketing data so every assistant learns the same records
   3. CExpose the systems through an open protocol that any compliant AI application can use as a client
   4. DCopy all records into a shared vector database that each assistant queries directly
   Show answer and explanationHide answer

   Correct answer: C

   MCP is an open standard, so one server implementation can serve any compliant host and avoid per-assistant integrations. Proprietary plugins duplicate effort, fine-tuning does not give live access to the systems, and a copied vector store goes stale and duplicates sensitive data.

5. Question 5Architecture & Components

   In an MCP deployment, which component creates and manages a dedicated client for each server connection?

   1. AEach individual MCP server
   2. BA separate relay process that the user installs to route every connection through one shared channel
   3. CThe language model's inference endpoint
   4. DThe host application
   Show answer and explanationHide answer

   Correct answer: D

   The host is the AI application that owns the user experience, and it creates one client per server it connects to. Servers, the inference endpoint, and a user-installed relay do not create those client connections in the defined architecture.