Power User's Guide to Large Language Models Analysis · Methodology · Cognitive Systems Power User's Guide to Large Language Models Beyond Prompt Engineering, Automation, and One-Shot Outputs This guide describes how to use large language models as reasoning environments, cognitive instruments, and components inside a long-term knowledge system. It is not about prompt tricks or generic productivity advice. Core Principle Prompts control outputs. Systems control thinking. Power users design workflows, not isolated prompts. § 01 Two Fundamental Modes of Work Almost all misuse comes from confusing these modes. Exploration Mode (High Entropy) The goal is to discover structure not yet fully articulated. Ideas are rough, ambiguity is allowed, contradictions are tolerated, and editorial friction is kept as low as possible. Input methods that work well: voice dictation, stream-of-consciousness...
The Savannah Protocol · Red‑Flag Framework (corrected footnotes) THE SAVANNAH PROTOCOL Absolute Red‑Flag Framework for Psychedelic Safety ⛔ ABSOLUTE RED FLAG — NON‑NEGOTIABLE — ABORT IMMEDIATELY (NON‑CLINICAL CONTEXT) If anyone agrees to sit for you while you are in an active major crisis, ABORT. No exceptions (outside a licensed, medically supervised program designed for acute care). This rule overrides: trust · reputation · warmth · experience · spiritual framing · your own desire for relief. A psychedelic session increases psychological permeability. A crisis already means reduced stability. Combining the two creates unbounded risk—especially without clinical screening, monitoring, and emergency capacity. Contemporary clinical protocols rely heavily on screening, preparation, monitoring, and exclusion criteria to reduce predictable harms. [1] ...