The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple sources including documentation search results, runtime diagnostic logs, and browser content.
Ingestion points: Content retrieved via the nextjs_docs tool and runtime resources such as nextjs://logs and nextjs://errors accessed through nextjs_call.
Boundary markers: The instructions do not define specific delimiters or security headers to separate ingested external data from the agent's core instructions.
Capability inventory: The agent has the ability to modify project files (e.g., updating dependencies or configuration during upgrades), execute browser scripts, and access sensitive environment data.
Sanitization: There is no mention of sanitization, filtering, or validation of the external content before it is processed by the agent.
[COMMAND_EXECUTION]: The browser_eval tool includes an evaluate action that allows the AI agent to execute arbitrary JavaScript code within a browser instance.
[DATA_EXFILTRATION]: The skill provides tools to access sensitive project information, including environment variables via the nextjs://env resource and development logs via nextjs://logs. Additionally, the browser_eval tool can be used to take screenshots and capture console logs from local or remote web pages.
[EXTERNAL_DOWNLOADS]: The skill instructs users to install the next-devtools-mcp package using npx commands from the npm registry.
[DYNAMIC_EXECUTION]: The browser_eval tool facilitates dynamic script execution by allowing the agent to run code directly in a browser environment to verify page behavior or test interactions.
[OBFUSCATION]: The pre-configured installation link for the Cursor editor contains a Base64-encoded config parameter that decodes to the installation command npx -y next-devtools-mcp@latest. This is a standard configuration pattern for the platform's installation process.

next-devtools-mcp-skill