nextjs-chatbot
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
- [SAFE]: The skill consists entirely of documentation, guidelines, and architectural patterns for developers. It does not include executable scripts, shell commands, or obfuscated code. Sources referenced, such as official SDK documentation and popular UI libraries, are well-known developer resources.
- [INDIRECT_PROMPT_INJECTION]: The documentation explicitly addresses the risks of processing untrusted data and provides a framework for mitigation.
- Ingestion points: The skill describes ingesting external web search results (web-search.md) and database retrieval data (search.md).
- Boundary markers: Recommends using grounded retrieval, source scope limits, and authoritative record validation (search.md).
- Capability inventory: Identifies the use of tools for operations like file deletion and side-effect execution (hitl.md, suggestions.md).
- Sanitization: Mandates the sanitization of rendered Markdown/HTML and the parameterization of database queries (tool-rendering.md, search.md).
- [DATA_EXFILTRATION]: The skill includes informative warnings regarding Server-Side Request Forgery (SSRF). It instructs developers to validate fetch destinations and bound request sizes to prevent unauthorized access to private network resources during web searches (web-search.md).
- [CREDENTIALS_UNSAFE]: The guidelines promote secure secret management, such as using environment variables for HMAC signing of tool approvals and advising against trusting client-side session IDs for authorization (hitl.md, persistence.md).
Audit Metadata