The Agent Skills Directory

[PROMPT_INJECTION]: The /project check command (defined in SKILL.md) audits existing project files for convention compliance, which creates an indirect prompt injection surface. Malicious instructions embedded in the audited source code could potentially influence the agent's behavior. (Ingestion points: The agent reads file content from directories such as app/Models/, app/Enums/, and routes/ during the audit process. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the auditing logic. Capability inventory: The skill has the capability to write arbitrary PHP files and Shell scripts to the local filesystem. Sanitization: The content of the audited files is not sanitized before processing by the agent.)
[COMMAND_EXECUTION]: The skill generates executable content from templates, including shell scripts and dynamic PHP loading logic. (The references/project-structure.md file provides templates for bin/setup.sh and bin/deploy.sh which execute system-level commands like composer install and php artisan migrate. The references/helper-conventions.md file includes a require_all_in helper function that uses glob and require to dynamically load PHP files from a directory.)
[EXTERNAL_DOWNLOADS]: The generated setup and deployment scripts include commands to download and install external dependencies from standard package registries (Composer and NPM), which is standard vendor functionality for Laravel projects.

project-laravel