web-search-valyu

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

• Third-party content exposure detected (high risk: 0.95). The skill’s runtime workflow calls Valyu endpoints like /v1/search, /v1/answer, and /v1/contents that fetch and extract readable text from arbitrary web URLs/sources provided at runtime (outsider-authored public web content), which is then blended into the agent’s LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill issues runtime API calls to https://api.gooseworks.ai (e.g., /v1/contents and /v1/search) which fetch arbitrary external pages (example shown: https://example.com/article) and inject that remote content into the agent's prompts/answers, so external URLs can directly control the agent's input.