scrapecreators-api

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

• Third-party content exposure detected (high risk: 0.85). The skill calls third-party ScrapeCreators API endpoints (e.g., /v1/.../profile, /v1/.../post/comments, /v1/.../transcript) using user-supplied URLs/handles/queries, and the runtime JSON responses can include outsider-authored free text (captions/comments/transcripts) that the agent would ingest into the LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent at runtime to fetch per-endpoint OpenAPI specs from https://docs.scrapecreators.com/{path}/openapi.json (and the full spec at https://docs.scrapecreators.com/openapi.json) to obtain parameter details and examples before making calls, so those external JSON docs would be fetched during runtime and directly control how the agent constructs requests (i.e., control its prompts/behavior).