solution-comparator

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content through the solutions and testCases input fields, which are directly used in the agent's workflow.
  • Ingestion points: The input schema in SKILL.md accepts code strings (solutions[].code) and arbitrary data in testCases.
  • Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from obeying instructions embedded within the provided code or test data.
  • Capability inventory: The skill is authorized to use Bash and Write tools, providing a path from data ingestion to command execution.
  • Sanitization: No sanitization, validation, or sandboxing of the input code is specified.
  • [DYNAMIC_EXECUTION]: The skill's primary function involves executing code provided as input at runtime.
  • The solutions parameter allows arbitrary code blocks to be passed to the agent. This code is intended to be executed via the Bash tool to perform benchmarking and correctness checks. If an attacker provides malicious code (e.g., shell commands disguised as algorithm solutions), it may be executed by the agent without sufficient oversight.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 08:23 PM