The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for processing untrusted HTTP request data such as headers, cookies, and URLs. It includes a mandatory evidence chain for this attack surface: ingestion occurs via request.headers and request.cookies as described in SKILL.md; boundary markers are maintained through recommended config.matcher rules; capabilities are inventoried including NextResponse redirects and header manipulation; and sanitization is addressed through explicit warnings against unsafe header forwarding and advocacy for route-level authorization.
[EXTERNAL_DOWNLOADS]: The content references the @vercel/functions package for geolocation and IP address retrieval. This is a trusted official library from a well-known service (Vercel) and represents standard industry practice for the platform.
[COMMAND_EXECUTION]: Includes standard instructions for performing framework migrations using official tools such as npx @next/codemod, which are verified developer utilities.

middleware-patterns