Generate Marketing Screenshots

================================================================================

🔴 VERDICT: HIGH

This skill presents a HIGH security risk primarily due to the powerful browser_evaluate command, which allows for arbitrary JavaScript execution within the browser context. While the skill's intended use of this command is benign (e.g., scrolling, dismissing popups), a sophisticated attacker could craft a prompt that manipulates the AI into generating and executing malicious JavaScript. This could lead to data exfiltration from the web pages being browsed, unauthorized actions within the browser, or other client-side attacks. Furthermore, the skill itself is installed from an external GitHub repository (anaghkanungo7/agent-skills) that is not on the list of trusted sources, introducing a supply chain vulnerability.

Total Findings: 3

🔴 HIGH Findings: • Indirect Prompt Injection / Arbitrary JavaScript Execution

• SKILL.md, Line 133: The browser_evaluate command allows the execution of arbitrary JavaScript within the browser context. If a malicious prompt can inject JavaScript into this command, it could lead to data exfiltration (e.g., document.cookie, local storage, form data) or other browser-based attacks on the navigated pages.

🟡 MEDIUM Findings: • Unverifiable Skill Dependency

• README.md, Line 40: The skill is installed from npx skills add anaghkanungo7/agent-skills/generate-marketing-screens. The source repository anaghkanungo7/agent-skills is not a trusted GitHub organization or specific repository. This means the skill's own code is sourced from an untrusted external location. • Indirect Prompt Injection / Malicious URL Navigation
• SKILL.md, Line 29: The skill navigates to user-provided URLs using browser_navigate. If the Playwright environment is not sufficiently sandboxed, navigating to a malicious URL could expose the underlying system or browser environment to attacks (e.g., drive-by downloads, browser exploits).

🔵 LOW Findings: • No low findings.

ℹ️ TRUSTED SOURCE References: • No trusted source references.

================================================================================