The token starts counting the moment you get it, so grab it when you're ready to work, not weeks before. And yes, a careless worker can burn through fetches, so point it only at what you need.
The crawl token is one of two ways into the free kit. One is the 206-page PDF you read yourself. The other is the token your AI worker uses to fetch every chapter, module spec and file straight from the server while it works. The kit itself is free. There's no paid tier behind it, nothing to unlock.
Here's how I'd keep the token from getting eaten:
- Decide what the worker fetches before you start it. Chapters it needs, module specs it needs, nothing else.
- Run it in a bounded spec so it does not loop back through the same files.
- Check the worker's output after the first pass, not at the end.
- Keep the 12-module spec handy at /kit-documentation/ so it has a map and stops guessing.
I don't publish an exact fetch count or a per-request breakdown. That detail is still being recorded, so I won't quote a number I can't back. Watch your own usage in the first session and you'll know your pace.
The cost of waiting is simple. The token is free now. Spend months deciding and you ship nothing, which is the only real bill here.
Get the kit, point the worker at one module, review it. That's the first evening.
Mouldi
