ubiquitous-language

Pass

Audited by Gen Agent Trust Hub on Jul 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill instructions are focused exclusively on text processing and local file creation for documentation purposes. No malicious patterns, obfuscation, or unauthorized access attempts were identified.- [DATA_EXPOSURE]: The skill writes information to a specific local file (UBIQUITOUS_LANGUAGE.md). This is the stated primary purpose of the skill and does not involve sensitive system directories or credentials.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (the conversation history) to generate its output. While this provides a surface for indirect prompt injection, the risk is negligible as the output is restricted to a documentation file and does not influence security-critical agent capabilities.
  • Ingestion points: Current conversation history.
  • Boundary markers: None explicitly defined for the conversation scan.
  • Capability inventory: Local file write to UBIQUITOUS_LANGUAGE.md.
  • Sanitization: None specified for the extracted terms.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 17, 2026, 02:40 AM
Security Audit — agent-trust-hub — ubiquitous-language