design-doc-review-loop
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates as an orchestrator for a design document review-revision cycle. It identifies target documents in a local directory (
docs/design/), invokes specialist subagents for critique and revision, and updates the document status based on the results. - [SAFE]: No external network operations, credential harvesting, or unauthorized privilege escalation attempts were detected. All file operations are scoped to the project's documentation files.
- [SAFE]: The use of subagents follows standard platform patterns for delegating specialized tasks. The skill includes a 'Delegation Requirement' section that instructs the agent to seek user confirmation if the intent to use subagents is not clearly established, which is a security best practice for autonomous orchestration.
- [SAFE]: Indirect Prompt Injection Surface: The skill processes untrusted content from design documents to determine the number of revision cycles needed. While this is an attack surface (Category 8), the skill's capabilities are limited to reading/writing documentation and coordinating other internal skills, with a hard limit of 5 iterations to prevent infinite looping. This does not pose a significant security risk in this context.
Audit Metadata