fal-ai-model-search

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

• Third-party content exposure detected (high risk: 0.80). The skill fetches and parses public content from fal.ai (e.g., search results at https://fal.ai/explore/search?q=... and model documentation via https://fal.ai/models/.../llms.txt), which are open third‑party pages the agent reads and interprets as part of its workflow and could contain untrusted/user-provided text enabling indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill fetches and parses the model documentation at runtime from the llms.txt endpoint (e.g. https://fal.ai/models///llms.txt), and uses that fetched content to drive prompts/integration and inject code examples, so this external URL directly controls the agent's behavior.