extension-ui
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local project files during its analysis workflow, creating an indirect prompt injection surface if the scanned files contain malicious instructions.
- Ingestion points: Processes
package.jsonand extension UI component files (popup, sidepanel, options, content scripts) inSKILL.md(Step 1 and Step 2). - Boundary markers: Absent; no explicit text boundaries or instructions to ignore potential commands within the analyzed code files are defined.
- Capability inventory: No shell command execution or network capabilities are active or enabled for this skill.
- Sanitization: Absent; the code is read directly for layout and accessibility analysis without a parsing filter.
Audit Metadata