bs-social-card
Fail
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: HIGHDATA_EXFILTRATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
- [DATA_EXFILTRATION]: The skill accepts user-provided local file paths or URLs for the LOGO_SRC and AVATAR_SRC placeholders. These values are inserted into HTML templates without validation. A malicious request could specify paths to sensitive system files such as SSH keys or environment configurations, causing the screenshot tool to render the file contents into the final image, potentially exposing private data.\n- [REMOTE_CODE_EXECUTION]: The skill dynamically constructs HTML files by interpolating user input into templates. While titles and subtitles are escaped, the lack of escaping on logo and avatar sources allows for HTML/JavaScript injection. A malicious URL could execute code within the local browser context to access local data or perform unauthorized actions.\n- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands via npx playwright. While intended for screenshotting, this involves running external CLI tools and requires software installation on the host system.\n- [EXTERNAL_DOWNLOADS]: The skill triggers the download and installation of the Chromium browser via the Playwright CLI, which is a significant external dependency fetched from a remote source.
Recommendations
- AI detected serious security threats
Audit Metadata