app-package

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation includes templates that reference external machine learning models and tokenizers from Hugging Face (huggingface.co). These are standard resources for Vespa's native embedding components and target a well-known service for model hosting.
  • [CREDENTIALS_UNSAFE]: The skill provides examples for configuring mTLS certificates (security/clients.pem) and access tokens (token id="my-read-token"). These are presented as placeholders and instructional examples for standard security configuration rather than hardcoded secrets.
  • [COMMAND_EXECUTION]: The instructions cover the use of the official Vespa CLI (vespa) for deploying and monitoring application packages. These commands are standard operations for managing the lifecycle of a Vespa application.
  • [INDIRECT_PROMPT_INJECTION]: The skill functions as a template generator for configuration files based on user requirements. While this creates a surface where user-provided data is interpolated into structured XML/YAML, the risk is mitigated by Vespa's own schema validation and the nature of the configuration engine.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 12:58 PM