write-prd

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user arguments and interpolates them into agent tasks and file system operations, creating a potential vector for injection.
  • Ingestion points: User-supplied product or feature names are captured via the $ARGUMENTS variable in SKILL.md.
  • Boundary markers: The skill does not implement delimiters or 'ignore embedded instructions' warnings when using the {PRODUCT} variable in task subjects or output filenames.
  • Capability inventory: The environment allows powerful tools including Bash, Write, Edit, and Agent delegation.
  • Sanitization: No input validation, character filtering, or escaping logic is present to ensure that the user-provided string does not contain malicious command sequences or path traversal patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 11:32 AM
Security Audit — agent-trust-hub — write-prd