playwright-codegen-stabilizer
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill guides the agent to capture full-page screenshots, HTML source code (via
page.content()), and accessibility snapshots as diagnostic artifacts during failures. While these files contain the data rendered on the target website, this behavior is a standard and necessary practice for the stated purpose of debugging and stabilizing web automation. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it requires the agent to read and analyze untrusted content from external websites (DOM, HTML, text) to propose script improvements. A malicious site could attempt to embed instructions within its markup to influence the agent's logic during the stabilization workflow. However, this is a general risk inherent to web-interactive tools and no specific malicious exploitation instructions are present in the skill itself.
- [COMMAND_EXECUTION]: The instructions involve the use of the Playwright CLI and browser interaction commands. These are used within the context of developer automation and do not involve unauthorized shell access or privilege escalation.
Audit Metadata