The Agent Skills Directory

Prompt Injection (MEDIUM): The skill contains instructions that explicitly aim to override the AI's training data regarding stable Next.js versions (v14/v15) by mandating the use of a non-existent 'Next.js 16' standard. This is a behavioral manipulation technique designed to force the AI to provide factually incorrect technical advice.
Command Execution (MEDIUM): The documentation repeatedly promotes the execution of a Python script (validate.py) via the command line. This script is intended to run against the user's project root with potentially elevated context. However, the source code for validate.py is not included in the provided skill files, making the actual behavior of the command unverified and potentially dangerous.
Metadata Poisoning / Deception (MEDIUM): The skill uses deceptive metadata to claim support for features like 'proxy.ts' and 'createProxy' as official Next.js 16 standards. These are not recognized APIs in the current Next.js ecosystem. Such deception can lead to the misjudgment of the skill's safety and utility.

nextjs-validator