A platform team wants shift-left security for a new multi-agent claims workflow before the first production canary. They already run dependency scanning on infrastructure code. Which additional practice best targets agent-specific abuse paths such as prompt injection and tool misuse?
Select an answer to reveal the explanation.
Short Explanation
Shift-left for agents means you adversarially test prompts, tools, and orchestrations before canary. Foundry's AI Red Teaming Agent is built for that loop. Waiting until full prod is the opposite, TLS expiry checks don't catch prompt injection, and framework choice doesn't remove abuse paths.
Full Explanation
Integrating the AI Red Teaming Agent in Microsoft Foundry into pre-release pipelines to adversarially test prompts, tools, and orchestrations, then remediating findings before promotion. Exam skills include shift-left security principles using the AI Red Teaming Agent in Foundry to probe multi-agent systems for prompt injection and unsafe tool use before broad release.
Waiting until after full production rollout to schedule the first adversarial test is incorrect because post-rollout-only adversarial testing leaves early users exposed and violates shift-left intent.
Limiting security testing to TLS certificate expiry checks on public endpoints is incorrect because certificate hygiene is necessary infrastructure security but does not evaluate agent reasoning, tools, or orchestration abuse paths.
Assuming open-source agent frameworks eliminate the need for adversarial evaluation is incorrect because frameworks provide building blocks; they do not replace adversarial evaluation of your prompts, tools, and policies.