international-ai-deploy

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted user input to generate local configuration files.
  • Ingestion points: User-provided values for api-key, baseurl, provider, and model-name are collected in SKILL.md.
  • Boundary markers: None provided; user inputs are directly interpolated into JSON templates (e.g., ${api-key}).
  • Capability inventory: The skill targets the local file system for configuration storage at .codebuddy/models.json and ~/.config/codebuddy/models.json (SKILL.md).
  • Sanitization: No validation, escaping, or filtering of user-provided content is specified before the configuration is generated or saved.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 05:44 PM