aios-long-running-harness

Pass

Audited by Gen Agent Trust Hub on Apr 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a documentation-only harness that provides operational checklists and execution patterns for AI agents. It does not include any scripts, binary files, or automated commands.
  • [SAFE]: Analysis of the instructions reveals no attempts at prompt injection, data exfiltration, or credential harvesting.
  • [SAFE]: References to platform-specific environment variables (e.g., AIOS_SUBAGENT_CLIENT) and CLI tools (e.g., codex-cli) are contextual configurations for the intended runtime environment and do not pose a security risk.
  • [SAFE]: The skill follows security best practices by recommending human-in-the-loop gates for sensitive operations like payments or policy-sensitive actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 15, 2026, 02:02 AM