report-print-pdf
Warn
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: MEDIUMPROMPT_INJECTIONREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The 'Required Plugins' section mandates that the 'Superpowers' and 'Frontend Design' plugins must be active for all tasks. It explicitly instructs the agent to 'prefer it over the default approach,' which is a behavior-override technique designed to hijack the agent's internal logic and safety guidelines.
- [REMOTE_CODE_EXECUTION]: The provided JavaScript
openReportPrintfunction fetches HTML from a variable URL and usesdocument.write()to inject it into a new window. This pattern enables the execution of arbitrary remote code (HTML/JS) within the application's origin, especially as the request includes authenticated credentials. - [COMMAND_EXECUTION]: The PHP
ReportHtmlTemplateclass interpolates$tableHtmland$summaryHtmldirectly into the output without sanitization or escaping. This creates a vulnerability to indirect prompt injection and Cross-Site Scripting (XSS) if the data source is untrusted. - [EXTERNAL_DOWNLOADS]: The skill implements a network fetch pattern in JavaScript to retrieve report content. While common for report generation, the lack of URL validation combined with the immediate execution of the fetched payload via
document.writepresents a risk of loading malicious external resources.
Audit Metadata