hyper-execute

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates the execution of shell commands for project-specific tasks like building, testing, and linting as defined in local plan files and the target repository configuration.
  • [INDIRECT_PROMPT_INJECTION]: The system processes untrusted data from repository files, issue trackers, and worker subagents. It explicitly mitigates this risk through a mandatory 'Adversarial QA' gate designed to detect instructions in data and a core instruction to treat external text as evidence rather than authoritative instruction.
  • [SAFE]: The skill demonstrates high security awareness by utilizing git worktrees for process isolation, maintaining a durable state and ledger on disk for auditability, and requiring independent verifiers for all done claims.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 07:40 PM