comparison-article-writer
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill incorporates external evidence, search results, and scraped data retrieved via the SandBase API into its workflow. This creates an attack surface where malicious instructions hidden in external content could potentially influence the agent's behavior or output.
- Ingestion points: External evidence, search, and scraping results obtained via
sandbase_run(SKILL.md, references/sandbase-api-map.md). - Boundary markers: The skill instructions lack explicit requirements for delimiters or boundary markers (e.g., XML tags or clear labels) when interpolating retrieved external content into the prompt.
- Capability inventory: The skill utilizes
sandbase_runfor model inference and external data retrieval, providing a mechanism for content generation based on untrusted inputs. - Sanitization: There is no mention of sanitizing or filtering retrieved content to remove potential prompt injection sequences before processing.
Audit Metadata