preview-testing
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes local shell commands such as
npm run test:e2eandnpx playwright testto facilitate automated testing and visual regression workflows. \n- [INDIRECT_PROMPT_INJECTION]: The skill interacts with and reads data from external web content, which presents a surface for potential injection of malicious instructions into the agent's context. \n - Ingestion points: Web page structure, accessibility trees, and console logs are ingested via
mcp__claude-in-chrome__read_pageandmcp__claude-in-chrome__read_console_messages(SKILL.md). \n - Boundary markers: The skill does not provide instructions or delimiters to the agent to distinguish between its primary instructions and the external web content it retrieves. \n
- Capability inventory: The skill includes access to local shell execution (
npm,npx) and direct control over a browser environment (mcp__claude-in-chrome__computer). \n - Sanitization: No sanitization, validation, or filtering is applied to the data retrieved from external preview URLs before it is processed by the agent.
Audit Metadata