patient-question-content

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to read and process untrusted external data from public sources such as Reddit comments and news search results.
  • Ingestion points: Data entering the context via reddit/posts/{id}/comments and news/search tools.
  • Boundary markers: No specific delimiters or instructions to ignore embedded commands within the fetched data are present.
  • Capability inventory: The skill's primary function is generating structured text (content plans); it does not appear to have access to dangerous system-level tools or file-write capabilities beyond its own context.
  • Sanitization: The instructions do not specify any validation or filtering for the content retrieved from external community sources.
  • [SAFE]: The skill references local configuration files (.agents/product-marketing.md and .claude/product-marketing.md) to establish context. This is standard practice for agents operating within a specific project environment and does not involve accessing sensitive system credentials or user data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 10:28 AM
Security Audit — agent-trust-hub — patient-question-content