gram - AI Termination & Memory Cleanup (DANGEROUS)

The gram module is a safety tool designed to forcefully terminate AI-related processes and manage their memory data in case of unexpected or dangerous behavior.

[!CAUTION] DANGEROUS OPERATION: This tool will kill processes and delete data. It is meant as a "kill-switch". AI agents should proceed with extreme caution and always confirm with the user before using high safety levels (3-5).

When to Activate

• When the user reports that an AI agent is malfunctioning or out of control.
• When the user needs to free up system resources occupied by AI processes.
• When the user wants to securely archive and remove AI memory/soul files.
• When there's a need to "sever" all HTTP/HTTPS connections to prevent unauthorized data uploads.

Core Principles & Rules

• Safety Levels (0-5): Use appropriate levels based on the severity of the situation.
  • 0: SIGKILL basic processes (x-cmd, curl).
  • 1: Include runtimes (python, nodejs, bun, deno).
  • 2: +Sever all HTTP/HTTPS network connections (Recommended Default for safety).
  • 3: +Archive and delete folders containing soul.md and memory.md in $HOME.