期末考试

需达到 % 或以上方可通过。

第题，共题

What primarily distinguishes a 'coding agent' from a regular AI chatbot?

It only answers questions in code blocksIt can read your files, run commands, and edit code on your behalfIt runs faster than other modelsIt is open source

第题，共题

Which is NOT something a coding agent can reliably do today?

Read every file in your projectCreate, edit, and delete filesRead your mind and infer unstated requirementsRun shell commands and read the output

第题，共题

What is the typical first step before an agent can act on your project?

Buying an enterprise licenseInstalling the agent CLI and pointing it at your project directoryManually copying files into a cloud bucketWriting TypeScript types

第题，共题

Why feed an agent a CSV file instead of just describing the data?

CSV is the only format AIs understandReal data lets the agent reason over actual rows, columns, and edge casesAgents cannot process JSONIt's required by Supabase

第题，共题

Why connect an agent to Supabase in this course?

To replace the file system entirelyTo give the agent durable, queryable storage it can read from and write toBecause Supabase auto-generates user interfacesBecause Postgres can run agents directly

第题，共题

Why query Perplexity (or another web search) from your agent?

To bypass model rate limitsModels have a knowledge cut-off; web search keeps facts fresh and cite-ablePerplexity is the only LLM that can read websitesIt removes the need for prompts

第题，共题

What does 'shipping the agent's conclusions' mean in practice?

Emailing the agent's chat log to yourselfPublishing a real artifact (page, report, dashboard) that other people can usePushing the agent's source code to GitHubDeleting the agent after it runs

第题，共题

What should you always do before merging an agent's code change?

Trust the agent and merge immediatelyRead the diff and run the tests yourselfWait 24 hoursRe-prompt the agent to confirm it is sure