chrome-devtools
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's configuration and documentation specify downloading the
chrome-devtools-mcppackage from the npm registry using thenpxutility. - [COMMAND_EXECUTION]: The skill uses
npxto execute the MCP server and provides a tool,evaluate_script, which allows for the execution of arbitrary JavaScript within the context of a web page. - [DATA_EXFILTRATION]: The skill includes tools to capture screenshots, page snapshots, and network request logs, which involve reading potentially sensitive data from the browser environment.
- [PROMPT_INJECTION]: The skill processes content from external websites, creating a surface for indirect prompt injection where instructions embedded in a web page could target the agent's browser control capabilities.
- Ingestion points: External web content retrieved via
navigate_pageandnew_page(SKILL.md). - Boundary markers: No explicit delimiters or instructions to ignore embedded directives are defined for the processed web data.
- Capability inventory: File system access for saving traces/snapshots, network monitoring, and browser script execution (
evaluate_script) (SKILL.md). - Sanitization: No content filtering or validation of the external page data is described in the provided files.
Audit Metadata