TokenPad

6 step-by-step guides

How to do the thing

Procedures rather than essays. Numbered steps, the mistakes that make each one fail, and a tool at the point where you would otherwise have to go looking for one.

  1. 01How do I build an AI agent?How to build an AI agentEight steps from empty file to a config you can ship, priced as you go.8 steps11 min
  2. 02How do I calculate how much an LLM API will cost?How to calculate LLM API costsSix steps to a number you can take to a planning meeting.6 steps8 min
  3. 03How do I write a good system prompt?How to write a system promptSeven steps, in the order that matters, with the cost of each decision.7 steps9 min
  4. 04How do I count the tokens in my prompt?How to count tokensExact counts, which encoding applies where, and the honest limits.5 steps6 min
  5. 05How do I create a skill for an AI assistant?How to create an AI skillFive steps, and the one mistake that makes skills never fire.5 steps7 min
  6. 06How do I reduce the number of tokens my prompts use?How to reduce token usageSeven changes, ordered by return. Most teams find 40% in the first three.7 steps8 min

Why these are separate from the guides

The guides explain why something is true — how tokenization works, why chat costs compound, what prompt caching actually saves. These answer a different question: what do I do, in what order, right now.

Both matter and they serve different moments. If you are trying to understand a bill, read a guide. If you are trying to ship something before lunch, start here and follow the steps.

Every step that needs a calculation links to one of the 102 tools, all free and all running in your browser.