Skills
Navigate
skills/akillness/oh-my-unity3d/llm-monitoring-dashboard/Gen Agent Trust Hub

llm-monitoring-dashboard

Fail

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONPERSISTENCECOMMAND_EXECUTIONDATA_EXFILTRATIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs the agent to download a shell script from a remote GitHub repository (nooscraft/tokuin) and pipe it directly to bash for installation. This occurs both in the main installation step and within the safety-guard.sh script.
  • [PERSISTENCE]: The skill automatically modifies the system's crontab to ensure that monitoring, reporting, and alerting scripts run at regular intervals (every 5 minutes, hourly, and weekly), maintaining execution across sessions.
  • [COMMAND_EXECUTION]: The skill uses chmod +x to make several generated scripts executable and then runs them within the shell environment.
  • [DATA_EXFILTRATION]: The skill is configured to send collected metrics and cost summaries to an external URL provided in the SLACK_WEBHOOK_URL environment variable using curl.
  • [DYNAMIC_EXECUTION]: The skill dynamically generates Python and shell scripts (such as categorize_prompt.py and collect-metrics.sh) using Here-Docs and executes them to process data.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted prompt data in the collect-metrics.sh script, passing it as an argument to the categorization script and the tokuin CLI without sanitization, validation, or protective boundary markers.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/nooscraft/tokuin/main/install.sh - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 6, 2026, 09:10 AM