song-form

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface through its template override mechanism, which allows files within the user's project directory to supersede internal songwriting logic. If project files at the override path are malicious, they could influence the agent's behavior.
  • Ingestion points: ${CLAUDE_PROJECT_DIR}/songwriting/templates/pat-pattison/ and $ARGUMENTS payloads.
  • Boundary markers: Not present.
  • Capability inventory: Reading local files and writing generated artifacts to disk.
  • Sanitization: No sanitization or validation of template content is specified.
  • [COMMAND_EXECUTION]: The skill instructs the agent to write generated songwriting artifacts to the local file system according to a persistence configuration. This capability allows the skill to modify local project files based on its output.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 07:55 PM
Security Audit — agent-trust-hub — song-form