skill-forge

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by reading and analyzing local SKILL.md files during the review process. An attacker could potentially embed malicious instructions within a skill file to manipulate the auditor agent's output or scoring logic.
  • Ingestion points: Processes existing SKILL.md files when executing the review command.
  • Boundary markers: Lacks explicit delimiters or instructions to ignore nested prompt content within the files being audited.
  • Capability inventory: File read access and structured text generation.
  • Sanitization: Does not implement specific validation or sanitization for the input data before processing it as part of its auditing logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 04:28 PM