perplexity-search

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection. It retrieves real-time data from the web and provides it to the agent's context. If the retrieved content contains adversarial instructions, it could influence the agent's logic.\n
  • Ingestion points: The search query and the resulting text retrieved from the OpenRouter API in scripts/perplexity_search.py.\n
  • Boundary markers: The skill lacks explicit delimiters or instructions to treat the external search results as untrusted data.\n
  • Capability inventory: The skill performs network operations to fetch external data and returns it to the agent, which could be used as input for subsequent tasks.\n
  • Sanitization: No sanitization or validation is applied to the retrieved search results before they are passed back to the agent.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with the OpenRouter API (openrouter.ai) and utilizes the litellm Python package. These external references target a well-known service provider and a standard library, which are appropriate for the skill's stated purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 11:30 PM
Security Audit — agent-trust-hub — perplexity-search