llm-models
Warn
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses deceptive metadata in its description and model table. It references non-existent or unreleased versions of language models, such as "Claude Opus 4.5", "Claude Sonnet 4.5", "Claude Haiku 4.5", and "Gemini 3 Pro". As current versions for these providers are 3.5 and 1.5/2.0 respectively, this misinformation is a form of metadata poisoning that could lead users to execute commands under false pretenses.
- [EXTERNAL_DOWNLOADS]: The skill documentation points to an external installation script hosted on GitHub (
https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md) and suggests installing further skills vianpx skills add. These downloads come from theinference-shorganization, which is not a pre-verified or trusted source. - [COMMAND_EXECUTION]: The skill relies on the execution of the
infshCLI tool. It includes alogincommand which handles authentication, potentially exposing API keys or tokens to a third-party binary.
Audit Metadata