product-feedback-sim

Warn

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill dynamically modifies the Python search path using sys.path.insert to load code from a location outside its own package directory (~/.claude/skills/persona-sim). This dynamic loading from a computed path is used to load and execute local modules.
  • [PROMPT_INJECTION]: The skill ingests untrusted user data into its simulation workflows, which can lead to indirect prompt injection.
  • Ingestion points: User-controlled values for target, goal, and candidates in the Python snippets.
  • Boundary markers: No delimiters or instructions are used to separate user data from simulation commands.
  • Capability inventory: The skill interacts with the sim_engine and sampler libraries to generate feedback and scores.
  • Sanitization: There is no visible sanitization or validation of the input strings before they are processed by the simulation engine.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 27, 2026, 07:09 AM
Security Audit — agent-trust-hub — product-feedback-sim