Skills
Navigate
skills/first-fluke/oh-my-agent/oma-design/Gen Agent Trust Hub

oma-design

Fail

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill uses the bunx utility to execute the getdesign@latest package from the public npm registry to list and fetch design templates.
  • [COMMAND_EXECUTION]: Technical resources describe the use of shell pipelines to process remote content, specifically piping curl output into python3 for JSON parsing and into tar for streaming file extraction.
  • [EXTERNAL_DOWNLOADS]: The skill fetches manifest data and template files from external domains including registry.npmjs.org and getdesign.md to provide design inspiration seeds.
  • [PERSISTENCE]: Instructions and commands are provided to modify global and workspace-level configuration files (such as mcp_config.json, .vscode/mcp.json, and ~/.gemini/config/mcp_config.json) to register and maintain MCP server connections across sessions.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external markdown templates from a community catalog. It includes specific mitigations such as framing instructions and hash verification to prevent instruction hijacking from these untrusted data sources.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 11, 2026, 04:38 AM