docker-sandbox

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements robust security hardening for Docker execution, including dropping all Linux capabilities (--cap-drop ALL), preventing new privileges (--security-opt no-new-privileges), and enforcing a read-only root filesystem.
  • [COMMAND_EXECUTION]: The primary function of the skill is to execute shell commands within a containerized boundary. This is intended to provide isolation for untrusted code and protect the host system from potential side effects.
  • [SAFE]: Networking is disabled by default (--network none) and write access to the workspace requires explicit user-provided flags, following the principle of least privilege.
  • [SAFE]: No instances of prompt injection, credential exposure, or obfuscation were detected in the skill instructions or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 12:56 PM