How LLMs actually read your prompt
Tokens, context windows and why long chats drift, in plain words.
read() →~/pheeniks/ai $ _
import { guides, tutorials } from "./ai";
Plain-language guides for getting better answers out of ChatGPT, Claude and other AI tools. Every guide shows the prompt I started with, what I changed, and why.
- summarize this
+ You're helping a busy manager.
+ Summarize in 5 bullets, plain words,
+ and flag anything that needs a decision.
every guide shows the diff like this
## 01 · guides/prompts
## 02 · guides/llm
same task, different models, honest results
## 03 · learning.log
[Text here: date] started AI agents + LLMs course
[Text here: date] wrote [Text here: first guide]
[Text here: date] tested [Text here: what you compared]
[Text here: date] next [Text here]
get new guides when they drop
[Text here: where people follow for new guides, e.g. a newsletter or YouTube.]