The Agent Skills Directory

[SAFE]: The skill provides instructional guidance for configuring metadata headers and DNS records. It includes proactive security advice based on OWASP ASI (Agent Security Issues) to prevent phishing and unauthorized access.
[INDIRECT_PROMPT_INJECTION]: The skill possesses the capability to read and modify codebase configuration files, which constitutes an attack surface for processing untrusted data from the local environment.
Ingestion points: Accesses project files via read_file and search_code tools mentioned in the SKILL.md metadata.
Boundary markers: The instructions specify strict RFC 8288 compliance and provide detailed validation checklists to ensure correct formatting and scoping.
Capability inventory: Operations are limited to standard file system actions such as write_file, edit_file, and create_file to update edge deployment configurations (e.g., Cloudflare _headers).
Sanitization: The skill mandates strict allowlisting of internal paths and relation types, and explicitly forbids the inclusion of third-party URLs in Link headers to prevent goal hijacking and redirection attacks.

configure-agent-headers