laravel-ai-sdk
Pass
Audited by Gen Agent Trust Hub on Mar 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill documents capabilities for agents to ingest untrusted data from external sources, creating a surface for indirect prompt injection.\n- Ingestion points: Instructions cover tools such as WebSearch, WebFetch, and FileSearch which ingest data from the public web and vector stores (rules/tool-provider.md).\n- Boundary markers: Prompting patterns in the guide lack explicit delimiters or instructions to the agent to ignore commands embedded in data attachments (rules/agent-prompting.md).\n- Capability inventory: Agents are equipped with custom tools capable of executing PHP code and performing file or media generation (rules/tool-create.md, rules/media-images.md).\n- Sanitization: The documentation does not provide guidance on sanitizing retrieved content for malicious instructions before the agent processes it.\n- [PROMPT_INJECTION]: The skill metadata identifies the author as 'Laravel Community' while the provided context identifies 'AsyrafHussin'. This discrepancy represents a metadata poisoning risk as it could mislead users or agents into assuming the skill is an official Laravel project.
Audit Metadata