skills/veniceai/skills/venice-models/Gen Agent Trust Hub

venice-models

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents how the agent should ingest and process external model metadata from API endpoints. This introduces a surface for indirect prompt injection if the model descriptions or trait names returned by the API were to contain malicious instructions designed to influence the agent's behavior.
  • Ingestion points: API endpoints defined in SKILL.md (/models, /models/traits).
  • Boundary markers: The skill does not define specific delimiters or warnings for the agent to ignore instructions embedded in the API response.
  • Capability inventory: The skill demonstrates using this data to programmatically select models, validate request parameters, and calculate token costs.
  • Sanitization: No explicit sanitization of text-based fields (like descriptions) is documented in the usage patterns.
  • [SAFE]: All network operations and URL references in the skill, such as https://api.venice.ai/api/v1/models, belong to the author's own infrastructure or well-known services like Hugging Face. The skill is purely informational and does not ship with any executable scripts, privilege escalation attempts, or persistence mechanisms.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 09:34 PM
Security Audit — agent-trust-hub — venice-models