The correct option is It constructs the prompt from the nearby code around the cursor including visible content comments and signatures in the current file.

In an IDE, Copilot forms completion prompts from the local editing context so it looks at the text around the caret in the active file. It uses visible code, comments, documentation strings, and function or method signatures to infer intent and suggest relevant completions. This approach keeps prompts focused and responsive while preserving privacy by limiting what is sent.

It performs a cloud semantic search over repository embeddings before every suggestion is incorrect because inline completions rely on the local editor context and are not driven by a repository wide embedding query before each suggestion.

It uploads the entire workspace to the model so it can use full project context is incorrect because Copilot does not transmit your whole project for each suggestion and instead sends only the necessary snippets from the active editing context to build the prompt.

It relies solely on the function name at the caret is incorrect because Copilot considers the surrounding code and comments in the current file and not just a single symbol name.

Full AWS Practitioner Certification Question