anthropic-sdk-pro

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a technical documentation resource for the Anthropic SDK. It contains no executable malicious code and follows security industry standards for AI integration.
  • [SAFE]: It provides explicit instructions for securing API keys, including preventing exposure via the NEXT_PUBLIC_ prefix and ensuring dangerouslyAllowBrowser is disabled in the SDK configuration.
  • [SAFE]: It implements defensive programming patterns, such as an input sanitization function that filters common prompt injection keywords (e.g., 'ignore', 'disregard') and recommends the use of XML tags to delineate user input from system instructions.
  • [SAFE]: The skill references only official and well-known dependencies, such as @anthropic-ai/sdk and the Vercel AI SDK, within its integration patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 01:20 AM