Skills
skills/omer-metin/skills-for-antigravity/prompt-caching/Gen Agent Trust Hub

prompt-caching

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill handles untrusted user data via variables like userQuery in patterns.md and CAGSystem.query. There are no boundary markers or sanitization steps before this data is interpolated into prompts.
  • Ingestion points: userQuery in patterns.md and CAGSystem.query.
  • Boundary markers: Absent in all code examples.
  • Capability inventory: Network operations to Anthropic API and Redis databases.
  • Sanitization: None detected in the provided patterns.
  • [External Downloads] (LOW): Implementation patterns reference external Node.js dependencies including @anthropic-ai/sdk and ioredis. While these are standard, they represent external code that must be verified.
  • [Credentials Unsafe] (INFO): The code snippets use process.env.REDIS_URL. While this is a recommended practice over hardcoding, it reminds users to protect their environment configurations from exposure.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 02:39 AM