base-ui-review
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
gitfor diff analysis and theghCLI for posting PR comments. These are well-known tools from a trusted organization (GitHub) used within their expected functional scope.- [INDIRECT_PROMPT_INJECTION]: The skill acts on untrusted data from code repositories. It mitigates injection risks through a strict "Subagent scope contract" that prevents the agent from executing code found in diffs. - Ingestion points: Code diffs and repository files (SKILL.md).
- Boundary markers: Explicit instructions in SKILL.md for subagents to treat files as data only.
- Capability inventory: File writing (via the
--fixflag) and network requests (via theghCLI). - Sanitization: Use of strict scoping and instruction-based constraints to prevent the agent from being misled by code content.
Audit Metadata