ai-sdk-ui-master

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill clearly integrates Mastra web/browsing tools (e.g., web-scraper-tool, browser-tool in src/components/ai-elements/tools and the Mastra tools list in mastra-reference.md) and the chat stream handling (app/chat/components/agent-tools.tsx and the API stream code using toAISdkStream with sendSources) renders and uses arbitrary public web/social content as tool outputs/sources that the agent reads and can influence subsequent behavior, so it consumes untrusted third‑party content that can affect actions.