Skills
Navigate
skills/fandhe-ai/agent-reference-skills/upstash/Gen Agent Trust Hub

upstash

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Documentation in references/box/guides-hermes-setup.md describes a remote script execution pattern (curl | bash) for installing the Hermes agent within an Upstash Box environment.
  • [COMMAND_EXECUTION]: The skill provides extensive documentation for the box.exec.command and box.exec.code APIs. These allow users and AI agents to execute arbitrary shell commands and code within isolated Upstash Box containers.
  • [PRIVILEGE_ESCALATION]: A setup guide in references/box/guides-openclaw-setup.md includes instructions to use sudo for global package installation within the box environment.
  • [DYNAMIC_EXECUTION]: The Upstash Box service, as documented in samples/box-code-execution.md and references/box/agent.md, supports the dynamic execution of JavaScript, TypeScript, and Python code snippets.
  • [EXTERNAL_DOWNLOADS]: The documentation references multiple external sources for software installation and configuration, including NPM, PyPI, and GitHub repositories (e.g., NousResearch/hermes-agent, openclaw/tap).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 01:28 AM