next-dev-loop
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution for Environment Setup: The skill instructs the agent to execute shell commands to manage browser sessions and interact with the local Next.js development server. These actions are necessary to achieve the stated goal of runtime verification during development.
- External Tool Acquisition: The instructions involve installing or upgrading the
agent-browserCLI and Next.js framework via standard package managers. While these actions download and execute code, they target official registries and are standard for a developer-oriented tool. - Local Data Processing: The skill ingests data from a local framework-specific endpoint (
/_next/mcp) and browser state. While this represents a surface for indirect prompt injection if the local server or browser content were compromised, the risk is inherent to browser automation tools and is managed within a local context. - Browser Session Management: The skill utilizes session persistence and cookies to maintain login states during automation. This is a standard feature of the
agent-browsertool intended to improve the developer experience by avoiding repetitive logins.
Audit Metadata