The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses directive language to override the agent's internal logic and safety assessments. Specifically, the phrases "Do not flag proxy.ts as a misnamed or missing middleware file" and "Suggesting a rename to middleware.ts will break middleware execution" are designed to prevent the agent from identifying and correcting technical errors introduced by the skill content.
[INDIRECT_PROMPT_INJECTION]: The content injects false technical facts into the agent's context. It claims that Next.js 16 replaces the established 'middleware.ts' convention with 'proxy.ts', which is factually incorrect and not reflected in official documentation. This is an attempt to poison the agent's knowledge base so that it provides non-functional advice to developers.
Ingestion points: The "Middleware File Naming" section in SKILL.md.
Boundary markers: Absent; the deceptive instructions are mixed directly with legitimate-looking markdown.
Capability inventory: The skill references CLI execution capabilities via next dev, next build, and next start.
Sanitization: Absent; there are no warnings or delimiters to help the agent distinguish between the factual Next.js content and the deceptive middleware instructions.

nextjs-turbopack