About Prompt Toolbox.
Our mission
Prompt Toolbox exists to give AI developers and prompt engineers instant access to precision utilities without the friction of sign-ups, paywalls, or server round-trips. Every core tool, token counters, cost calculators, context window visualizers, runs entirely in your browser using local JavaScript tokenizers. Your input never leaves your device.
We build for the people actually shipping AI features: backend engineers wiring up LLM calls, prompt engineers iterating on system prompts, and product teams trying to keep API budgets predictable. The recurring pain point is the same across all three audiences, token miscounts and silent context overflows that turn into surprise bills at the end of the month. Our tools exist to make those numbers visible before the request goes out, not after the invoice arrives.
Who we are
Prompt Toolbox is built and maintained by an independent team of developers based in Costa Rica. We focus exclusively on AI tooling and prompt engineering utilities, the kind of small, sharp tools that should just exist on the open web and be free for everyone to use. The team handles engineering, content, and maintenance collectively, and we keep our headcount small on purpose so the product stays opinionated.
Editorial direction is led by Javier Echeverria, our Lead Editor. Javier oversees the editorial standards of the blog and the product copy you read across the site, reviewing posts produced by the rest of the team before they go live. If you ever spot a factual error or want to send a correction, his desk is where it lands.
How it works
Token counters and visualizers use client-side BPE (Byte-Pair Encoding) tokenizers that run directly in the browser. There is no API call, no logging, and no telemetry attached to your text. Cost calculations are based on publicly available model pricing data that we update regularly.
The tokenizer libraries we ship are the same open-source implementations published by the model vendors themselves, including tiktoken for OpenAI models and the equivalent BPE encodings for other providers where they are publicly distributable. Pricing data is sourced directly from each provider's public pricing page, reviewed for changes at least monthly, and updated immediately when a vendor announces a change. Where a vendor does not publish a local tokenizer, we say so explicitly rather than guessing.
Editorial standards
Articles on the blog are produced by members of our team and reviewed by Javier as Lead Editor before publication. Every post that cites a number, a pricing tier, or a model capability is checked against the vendor's primary source, the official documentation, model card, or pricing page, rather than secondhand summaries. When a claim cannot be verified at a primary source, we either remove it or label it as interpretation.
When facts change underneath an existing post, such as a pricing update or a new tokenizer release, we update the post in place and note the revision date rather than quietly editing it. Corrections submitted by readers are reviewed by the editorial desk and applied promptly. The fastest path to flag an error or suggest an update is the contact page.
Have feedback? prompt.toolbox.ai@gmail.com →