skills/nousresearch/hermes-agent/grok/Gen Agent Trust Hub

grok

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill documents an installation procedure using a shell script from the official x.ai domain (curl -fsSL https://x.ai/cli/install.sh | bash).
  • [EXTERNAL_DOWNLOADS]: Recommends installing the @xai-official/grok package from the official NPM registry.
  • [COMMAND_EXECUTION]: Uses the terminal tool to execute the grok CLI for autonomous coding tasks and tmux for persistent interactive sessions.
  • [CREDENTIALS_UNSAFE]: The skill instructs the agent to verify authentication status by checking for the presence of the ~/.grok/auth.json file, which contains cached OAuth tokens.
  • [PROMPT_INJECTION]: The orchestrated agent processes external project data including CLAUDE.md files and git diffs, which acts as a surface for indirect prompt injection. 1. Ingestion points: Local project files, CLAUDE.md, and git diffs. 2. Boundary markers: None specified in the instructions. 3. Capability inventory: Terminal tool for shell execution, file manipulation, and subagent spawning. 4. Sanitization: No explicit content filtering mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 03:31 AM
Security Audit — agent-trust-hub — grok