vote-predict

Warn

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The Python code uses sys.path.insert combined with __import__('pathlib') to dynamically load and execute modules from a hidden directory in the user's home folder (~/.claude/skills/persona-sim). This allows the skill to run logic from a path that is not part of the skill's own codebase.
  • [PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection:
  • Ingestion points: User-provided input (question and options) is directly interpolated into the prompt in the ask function in SKILL.md.
  • Boundary markers: The prompt template lacks delimiters or specific instructions for the agent to ignore instructions embedded within the user data.
  • Capability inventory: The skill utilizes generate for LLM tasks and modifies the system path to access local library files.
  • Sanitization: There is no evidence of input validation or escaping before the data is passed to the LLM.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 27, 2026, 07:09 AM
Security Audit — agent-trust-hub — vote-predict