I discovered a tool called Runprompt that lets you chain LLM prompts together like standard Unix pipes, turning complex AI workflows into simple terminal commands.