The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references an external documentation page at https://jeffallan.github.io/claude-skills/skills/backend/laravel-specialist/. Automated scans (URLite) have flagged this URL as malicious. Although this appears to be the author's official documentation site, the alert should be reviewed before interaction. Additionally, the skill's main instruction file (SKILL.md) was flagged by automated file reputation scanners, though manual inspection of its markdown content does not reveal immediate malicious patterns.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute several shell commands within the project environment using the Laravel Artisan CLI. This includes commands like php artisan migrate:status, php artisan route:list, php artisan queue:work, and php artisan test in SKILL.md. It also suggests running the ./vendor/bin/pint linter. While these are standard tools for Laravel developers, they provide a direct mechanism for executing arbitrary commands if the agent is manipulated by malicious input.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes user requirements and external guidance to generate code and perform system operations.
Ingestion points: User-provided requirements for application features, architecture designs, and the skill's own reference documentation in SKILL.md.
Boundary markers: The instructions lack explicit delimiters or warnings to isolate user-provided data from the agent's core instructions, which could allow embedded malicious prompts to be executed.
Capability inventory: The skill has the capability to execute shell commands via the php artisan CLI and perform file system modifications (implied by the task of creating models, controllers, and tests) in SKILL.md.
Sanitization: There is no evidence of input validation or sanitization routines for the data ingested during the workflow.

laravel-specialist