agent-design
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill functions primarily as a design document and configuration specification for project-scoped advisor agents. It defines a YAML-based schema for setting up advisor personas without including any executable malicious logic.
- [COMMAND_EXECUTION]: The documentation references local maintenance scripts, such as
scripts/bridge_agents.pyandscripts/init.sh, which are used to synchronize agent configurations with the host environment (Claude Code and Codex). These are internal project tools and do not involve remote code execution or unauthorized command injection. - [EXTERNAL_DOWNLOADS]: No external URLs, remote script downloads, or unverifiable third-party dependencies were found in the skill content.
- [PROMPT_INJECTION]: The skill focuses on defining specialized value systems for sub-agents (e.g., readability over cleverness) and does not contain instructions aimed at bypassing the primary agent's safety filters or operational constraints.
Audit Metadata