huggingface-best

Fail

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: HIGHCREDENTIALS_UNSAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill instructions include commands that read the Hugging Face API token from the sensitive file path ~/.cache/huggingface/token.\n- [COMMAND_EXECUTION]: The skill relies on shell command execution via curl, jq, and the hf CLI to interact with external APIs and local files, including the use of command substitution to inject credentials into requests.\n- [DATA_EXFILTRATION]: The skill transmits the sensitive Hugging Face authentication token to huggingface.co. While this targets the official domain of a well-known service for intended functionality, it involves the automated transmission of local credentials.\n- [PROMPT_INJECTION]: The skill contains potential vulnerability surfaces for deception and indirect injection:\n
  • The skill's metadata claims it is an 'official' skill from the 'huggingface/skills' repository, which contradicts the actual authorship information provided in the context, posing a risk of deception regarding its origin and safety.\n
  • The skill ingests untrusted data from external leaderboard and dataset APIs without explicit boundary markers or sanitization of the content before it is processed into the agent's context.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 7, 2026, 08:24 PM
Security Audit — agent-trust-hub — huggingface-best