generate-laravel-project-overview

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it interprets untrusted project data to generate instructions for downstream agent tasks.
  • Ingestion points: The agent is directed to analyze the 'codebase' in SKILL.md.
  • Boundary markers: The prompt lacks delimiters or specific instructions to ignore malicious directives found within code comments, strings, or documentation in the analyzed files.
  • Capability inventory: The agent possesses the capability to write to the file system (specifically .ai/guidelines/project.md), which is explicitly used to guide other AI coding agents.
  • Sanitization: No sanitization, filtering, or validation steps are defined to prevent external content from influencing the generated guidelines.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 09:55 AM