plain-english
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided prose and possesses file-editing capabilities, creating a surface for indirect prompt injection.
- Ingestion points: The skill accepts user-pasted text for auditing and rewriting, and reads local files when invoked in 'Edit' mode (referenced in
SKILL.md). - Boundary markers: The instructions do not define clear delimiters or specific instructions for the agent to treat input as data and ignore any embedded directives.
- Capability inventory: In 'Edit' mode, the agent is instructed to use file-editing tools to modify text in place (
SKILL.md). - Sanitization: There is no evidence of input validation or sanitization to prevent instructions within the processed text from affecting the agent's execution context.
Audit Metadata