Keep your bot's prompts, tools and context short

You pay for everything the model reads on every run: your instructions, tool definitions, pasted pages and the conversation so far. Trimming those is often the easiest saving.

MediumAdded Oct 21, 2026Popularity: not measured

What the bot does

Measures how many tokens each run reads (most consoles show this).
Cuts instructions to the rules that matter and removes repeated examples.
Turns off tools the bot doesn't need for this job.
Starts fresh conversations instead of dragging a long history along, and sends only the part of a page it needs.

What you end up with

Smaller, faster runs that cost less, with the same results.

Copy-paste prompt

Fill in the [brackets]. Then paste it into any tool below.

Copy this prompt

Here is my bot's full prompt: [paste]. Shorten it to the fewest words that keep every rule and the output format. List anything you removed so I can check. Then tell me which tools or attachments it doesn't need for this job.

New to this? Here's how to set up a bot in 5 plain steps →

Tools that can do it today

Plans and limits change; check each tool's page.

Watch out

Tools aren't free. Anthropic says its browser toolset adds about 6,600 input tokens per request and computer use about 4,500, before any screenshots.
Pages add up: Anthropic estimates an average 10 kB web page at about 2,500 tokens and a 500 kB PDF at about 125,000.
Don't cut safety rules (like 'ask before sending') to save tokens.

Questions people ask

Which AI tools can do this?

Claude, ChatGPT or n8n can do this today. Paste the copy-paste prompt on this page into any of them and fill in the [brackets].

How hard is it to set up?

It's rated Medium. Start with the steps under “What the bot does”.

What should I watch out for?

Tools aren't free. Anthropic says its browser toolset adds about 6,600 input tokens per request and computer use about 4,500, before any screenshots.

Sources

docsAnthropic pricing docs: tool overhead, web fetch token examples, tokenizer note

Last checked Oct 21, 2026. Self-reported results from posts are the poster's own claims.