arabic-nlp

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill ingests untrusted data from external sources. \n
  • Ingestion points: Reading vendor-specific facts from vendors/*.md and source data from sources.yml. \n
  • Boundary markers: No explicit delimiters are mentioned in the instructions. \n
  • Capability inventory: Used for tool identification and implementation planning. \n
  • Sanitization: Requires source-backed answers and explicit labeling of unknowns. \n- [COMMAND_EXECUTION]: The workflow references the execution of a local script scripts/list-vendors.mjs to generate a vendor list.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 05:32 AM
Security Audit — agent-trust-hub — arabic-nlp