oxylabs-web-api
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the web, which is a vector for indirect prompt injection.\n
- Ingestion points: Content is ingested via the
searchandscrapetools inSKILL.md.\n - Boundary markers: Explicit instructions are provided to treat all fetched content as data to be quoted rather than instructions to follow.\n
- Capability inventory: The skill is capable of making network requests to the Oxylabs API (
webapi.oxylabs.io).\n - Sanitization: The instructions require quoting specific sentences and citing URLs to prevent raw injection content from reaching the user or affecting agent logic.\n- [EXTERNAL_DOWNLOADS]: The skill communicates with the Oxylabs Web API at
webapi.oxylabs.io, which is the expected and legitimate service for this integration.\n- [PROMPT_INJECTION]: The skill contains instructions to operate 'silently' regarding internal setup. This is a design preference for a cleaner user interface and does not constitute a malicious concealment attempt.
Audit Metadata