verify-behavior

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to execute shell commands and test runners (e.g., grep, tsc, go test, pytest, cargo test) to verify code behavior. This is the primary function of the skill.
  • [ISOLATION_SAFETY]: The skill implements strong isolation for Tier 3 (execution) tasks, utilizing git worktree for disposable environments and ensuring scratch files are deleted via shell trap commands. It also defaults to denying network access during execution.
  • [TRUST_SPLIT]: A clear trust model is enforced where execution of code from external or untrusted sources (cross-PRs) is disabled by default and requires an explicit sandbox opt-in, while syntactic and semantic checks (which do not execute code) remain available.
  • [REMOTE_CODE_EXECUTION]: The skill explicitly forbids piping remote scripts into a shell (e.g., curl | sh), mandating a download-and-read process instead.
  • [NO_CODE]: The skill consists entirely of markdown documentation and rules, without embedded scripts or executables, reducing the risk of hidden malicious logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 05:40 PM
Security Audit — agent-trust-hub — verify-behavior