chrome-automation

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes pkill -f chrome and pkill -f chromium (via shell or child_process.execSync) to manage system resources. These commands are documented as mandatory cleanup steps to prevent memory leaks and CPU exhaustion after automation tasks complete.
  • [EXTERNAL_DOWNLOADS]: The skill references several well-known and trusted automation libraries, including Puppeteer, Playwright, Selenium, and Cypress. These are standard tools for the skill's stated purpose of web automation and testing.
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection because it processes content from external URLs.
  • Ingestion points: page.goto(url) and data extraction via page.$$eval in SKILL.md.
  • Boundary markers: Absent.
  • Capability inventory: Shell command execution (pkill), file writing (screenshot, pdf, trace.json), and network navigation.
  • Sanitization: Not specified in the provided templates. This is a common characteristic of web scraping tools and does not escalate the verdict for this use case.
  • [SAFE]: All file operations (saving screenshots and PDFs) and network requests are consistent with the intended functionality of a browser automation tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 04:03 AM