Use coding agents for application development
Free Google Cloud Professional Agentic Architect practice — 6 questions on Use coding agents for application development, with explanations. No sign-up. Full 12-question mixed test →
A platform team wants Gemini Code Assist to automatically review generated code for security issues as part of their Cloud Build pipeline before any pull request can be merged. Which design BEST balances speed and governance?
AI code review accelerates issue detection but does not eliminate the need for human accountability; Google's responsible-use guidance for agentic coding tools requires a human-in-the-loop approval gate before code generated or reviewed by an AI tool is merged into production branches.
An enterprise needs its coding assistant to generate completions and chat responses that are grounded in the organization's private internal libraries and coding conventions, pulled directly from their internal source repositories. Which capability enables this?
Gemini Code Assist Enterprise includes a code customization capability that connects to an organization's private source repositories (via Developer Connect) so completions and chat are grounded in internal libraries, patterns, and conventions — a feature not available in the Standard tier.
A developer needs to use Gemini Code Assist inside their IDE within a Google Cloud project. Which IAM role must be granted to their identity to enable this access?
roles/cloudaicompanion.user is the specific predefined IAM role required for a principal to use Gemini Code Assist, and the Cloud AI Companion API must also be enabled on the project.
A coding agent suggests adding an import for a third-party package that sounds plausible but does not actually exist in any public registry — a known risk sometimes called package or dependency hallucination. What is the BEST mitigation an architecture team should enforce?
Generative coding agents can hallucinate plausible-but-nonexistent package names, which attackers can exploit by publishing malicious packages under those names (dependency/slopsquatting attacks). The architectural control is to route all dependency resolution through a curated Artifact Registry allowlist so unverified or hallucinated packages cannot be installed.
A developer wants to use Gemini CLI to autonomously refactor code across many files in a large repository with minimal manual intervention. Which configuration choice provides the BEST balance between development speed and safety for this production codebase?
For multi-file autonomous changes in a production codebase, the default interactive/checkpoint mode — which pauses for human confirmation before each file write or tool action — is the appropriate control, preserving developer oversight while still leveraging automation for the bulk of the work.
An architect is deciding how to add a code-generation capability to an organization's software delivery process. The requirement is purely to speed up how human developers write and review code during the SDLC; the capability does not need to be invoked at runtime by an application or orchestrated alongside other production agents. What is the MOST appropriate architectural choice?
When the requirement is accelerating human developers during the SDLC rather than providing a capability invoked by the running application, the correct architectural fit is an existing developer-tooling product like Gemini Code Assist integrated into the IDE/CI pipeline — not a custom runtime agent built and orchestrated as part of the production multi-agent system.
Ready for the real thing?
The full course has two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed answer explanations.
Start my full course on Udemy →