The Agent Skills Directory

[SAFE]: The skill provides comprehensive documentation and best practices for writing feature, unit, and browser tests with Pest v4. Analysis of SKILL.md and the accompanying reference files (core.md, browser.md, laravel.md) shows that all provided code snippets and commands are standard for Laravel development. There are no instances of obfuscation, hardcoded credentials, or suspicious remote code execution patterns.
[INDIRECT_PROMPT_INJECTION]: The skill focuses on generating and running tests for local project code, which creates an inherent attack surface for indirect prompt injection if the codebase being analyzed is untrusted.
Ingestion points: Local project source code and test files located in the tests/ directory as described in SKILL.md.
Boundary markers: Absent; the skill does not specify any isolation or delimiting markers when the agent reads untrusted application code.
Capability inventory: The skill suggests using the php artisan test command (referenced in SKILL.md), which executes the local application code and test logic.
Sanitization: Absent; no sanitization or verification of the project's code integrity is performed before the agent suggests or executes tests.

pest-testing