preview-verdicts

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted entries from the repository to generate an interactive review page.
  • Ingestion points: Entry data such as names, URLs, and reasons are sourced directly from the project repository and processed by the agent.
  • Boundary markers: Absent; the instructions do not include boundary markers or warnings to the agent to ignore potentially malicious content within the repository entries.
  • Capability inventory: The skill is capable of writing files to the local file system (tmp/ directory), executing shell commands (specifically the 'open' command), and modifying the README.md file.
  • Sanitization: Absent; the template.html logic uses direct string interpolation to render project data into the DOM, creating a Cross-Site Scripting (XSS) vulnerability surface where malicious repository content could execute JavaScript in the maintainer's browser.
  • [DYNAMIC_EXECUTION]: The skill dynamically assembles an HTML file containing JavaScript by replacing placeholders in a template with data fetched at runtime from the repository.
  • [COMMAND_EXECUTION]: The skill executes the system 'open' command to launch the generated HTML preview file.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 04:51 PM
Security Audit — agent-trust-hub — preview-verdicts