perplexity-data-handling

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements proactive security measures through a sanitizeQuery function that redacts PII (email, phone, SSN, tokens) before data is transmitted to the external Perplexity infrastructure.
  • [SAFE]: It includes citation validation logic that ensures third-party URLs returned by the search engine are using standard web protocols and are correctly formatted, mitigating risks from malformed or malicious links.
  • [PROMPT_INJECTION]: The skill processes data from an external search provider (Perplexity), which is a standard ingestion point for indirect prompt injection. However, the risk is mitigated by the skill's primary purpose of implementing data handling best practices.
  • Ingestion points: SKILL.md (Step 3: response.choices[0].message.content)
  • Boundary markers: Absent for the primary answer text
  • Capability inventory: allowed-tools: Read, Write, Edit
  • Sanitization: Robust citation validation and deduplication are implemented, though the main answer text is not explicitly escaped.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 07:55 AM
Security Audit — agent-trust-hub — perplexity-data-handling