clade-reliability-patterns

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill demonstrates code patterns that interpolate untrusted external data directly into model prompts without using boundary markers or sanitization logic.
  • Ingestion points: The userInput parameter in the getResponse function located in SKILL.md is passed directly into the message content array.
  • Boundary markers: The template lacks delimiters (like XML tags or triple quotes) or specific instructions to help the model distinguish between developer instructions and user-provided content.
  • Capability inventory: The scripts demonstrate usage of the Anthropic SDK messages.create method to perform network requests to an LLM provider.
  • Sanitization: The provided code does not include escaping or validation logic to mitigate malicious content within the userInput variable.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 01:02 PM
Security Audit — agent-trust-hub — clade-reliability-patterns