huggingface
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates communication with remote endpoints at
unbrowse.ai(the author's domain) to fetch data fromhuggingface.co. This is the core intended functionality for the MCP server and API. - [COMMAND_EXECUTION]: The documentation includes shell commands for the
claudeandcodexCLI tools to register the remote MCP server. These are standard configuration commands for the Model Context Protocol. - [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by processing external data from Hugging Face and allowing natural language task execution via the
unbrowse.runtool. However, the skill provides specific instructions to the agent to mitigate this risk, such as reporting only verified results and requiring explicit user approval before performing actions like posting, buying, or deleting data.
Audit Metadata