The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements a multi-agent workflow for exploration and validation without any instructions designed to override safety filters or manipulate the agent's core behavior. Instructions focus purely on pedagogical and implementation tasks.- [DATA_EXFILTRATION]: No hardcoded credentials or sensitive file harvesting patterns are present. The skill correctly directs users to manage secrets via environment variables and uses placeholders for external API configurations.- [OBFUSCATION]: Analysis of all 22 files revealed no hidden content, encoded strings, or homoglyph attacks. All documentation and templates are presented in clear, human-readable markdown and PHP.- [REMOTE_CODE_EXECUTION]: Installation instructions utilize standard package management tools (Composer and NPM) directed toward official first-party libraries such as Sanctum and Passport. No suspicious remote script piping or unverifiable downloads were found.- [INDIRECT_PROMPT_INJECTION]: While the skill interacts with project codebases, it provides secure-by-default templates and mandates a validation step using a specialized 'sniper' agent to ensure the correctness and security of generated code.- [COMMAND_EXECUTION]: Shell commands provided in the templates are standard Laravel developer operations (artisan commands, migration execution) and are intended for the developer to execute manually during setup.

laravel-auth