Module quiz

17 questions · pass bar 80% · retry as often as you like — your best score counts toward the gate.

  1. 1.In the MCP architecture, where do API credentials live, and why there?
  2. 2.Tools, resources, and prompts — who invokes each?
  3. 3.How does an MCP session begin?
  4. 4.Your API has 12 REST endpoints. The straightforward MCP server mirrors each as a tool. Why is this usually wrong, and what's better?
  5. 5.Why are tool descriptions 'prompts in disguise'?
  6. 6.A tool can match 200k tokens of results. What's the right response strategy?
  7. 7.How should a tool signal errors, and why not just raise exceptions?
  8. 8.Name the sandbox property and the attack it prevents. Which pairing is correct?
  9. 9.What problem does MCP solve versus A2A?
  10. 10.When should you choose stdio versus streamable HTTP transport?
  11. 11.What are the three layers of testing for an MCP server?
  12. 12.A model keeps calling your destructive delete_project tool during exploratory questions. Which defense is the right FIRST layer, within your server?
  13. 13.Your team just adopted MCP for a new agent product. A skeptical engineer says 'great, now our tools are automatically good, our context usage is handled, and third-party servers are safe to connect.' What's wrong with that claim?
  14. 14.You connect five MCP servers averaging eight tools each, and the agent's tool selection gets measurably worse — even though every individual tool is well-designed. What's happening, and what's the fix?
  15. 15.What is the 'confused deputy' problem as it applies to an MCP server, and why does it matter even if every user of your agent is fully trusted?
  16. 16.What is the 'lethal trifecta,' and why does a well-connected MCP host assemble it easily?
  17. 17.Why is connecting to a third-party MCP server a different — and larger — risk than adding a typical third-party library dependency?