software-engineering-flow

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists of purely instructional markdown text intended to guide task routing. There are no executable scripts, system commands, or file system operations defined within the skill.
  • [PROMPT_INJECTION]: No evidence of malicious instructions or jailbreak attempts. On the contrary, the skill explicitly lists 'Letting skills override direct user instructions, safety, or repo-local AGENTS.md' as a failure mode, which reinforces standard safety protocols.
  • [EXTERNAL_DOWNLOADS]: The skill does not reference external URLs, remote scripts, or third-party repositories. All references appear to be internal to the agent's skill environment.
  • [REMOTE_CODE_EXECUTION]: No patterns of dynamic code loading, subprocess execution, or remote command execution were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 08:57 AM
Security Audit — agent-trust-hub — software-engineering-flow