skills/hemsoft/public-skills/verbiage/Gen Agent Trust Hub

verbiage

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill provides a mechanism to store untrusted user data in a repository file which may influence downstream agent reasoning. ● Ingestion points: User-provided term definitions in SKILL.md actions. ● Boundary markers: None mentioned for delimiting user content. ● Capability inventory: File system read and write access restricted to the project glossary file (VERBIAGE.md). ● Sanitization: No input validation or escaping of markdown content is performed before writing to the file.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 08:45 AM