product-self-knowledge

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches product information and documentation maps from official Anthropic domains, including docs.claude.com and docs.anthropic.com. These are trusted resources used to provide accurate product capabilities.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: The agent is instructed to browse external content from docs.claude.com and support.claude.com.
  • Boundary markers: The instructions focus on factual accuracy and directing users to official URLs.
  • Capability inventory: The skill possesses no capabilities for code execution, file system access, or network operations beyond reading documentation.
  • Sanitization: Content is retrieved from trusted, vendor-controlled documentation sites.
  • [SAFE]: No obfuscation, persistence mechanisms, or privilege escalation patterns were detected. The skill is purely instructional and relies on trusted official sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 09:42 AM