10Design a customer-support agent with order-status and returns tools. How do you keep it safe and useful?▼medium★ EssentialSierraDecagonOpenAI1 repliesunlockedSierra's signature design exercise. The grading hinges on three things most candidates underweight: tool API shape, the read/write trust boundary, and eval cases written before the agent. Walkthrough inside.Open full answer →
16A customer wants to expose internal APIs to a model via MCP. Design it safely: auth, least privilege, versioning.▼mediumAnthropicOpenAIGlean1 replies○ sign inAn Anthropic-signature integration question. The hidden rubric: tools are a curated product surface, not a proxy for your OpenAPI spec, plus the auth model and confused-deputy trap most candidates miss.Open full answer →
29Your agent reads emails and can issue refunds. Harden it against prompt injection.▼hardAnthropicOpenAISierra1 replies◆ premiumUntrusted content plus write-capable tools is the lethal combination. Prompt-level defenses alone fail this question; the winning answer is an architecture where injection succeeding still can't move money.Open full answer →