venice-chat
Pass
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The documentation includes example
curlcommands for interacting with the Venice AI API. These commands target the official vendor domain (api.venice.ai) and use standard environment variables for authentication.\n- [PROMPT_INJECTION]: The skill describes an indirect prompt injection attack surface by documenting how to enable features likeenable_web_searchandenable_web_scraping. These features allow the agent to process untrusted data from the open web and search results.\n - Ingestion points: External content retrieved via web search and web scraping, as well as remote image and video URLs provided in message content.\n
- Boundary markers: None identified in the skill instructions.\n
- Capability inventory: The API facilitates network data retrieval and multimodal data processing.\n
- Sanitization: Not specified in the documentation; the skill relies on the underlying model's safety guardrails.
Audit Metadata