Token counter
Token counter counts tokens with the GPT-4 or GPT-4o tokenizer and shows word and character counts. It is exact for those OpenAI models and an estimate for Claude, Gemini, and Llama, so leave headroom under any hard limit.
← All tools
AI Developer
Count tokens, estimate LLM cost, and build prompts, rules files, and configs for AI tools. Everything runs in the browser and nothing is sent to a model.
Runs in your browser · Nothing uploaded
- Token counterEstimate how many tokens a prompt uses for common LLM tokenizers.Runs in your browser · Nothing uploaded
- Context window checkerSee how much of a model's context window your text fills.Runs in your browser · Nothing uploaded
- LLM cost calculatorEstimate API cost from input and output tokens and per-million pricing.Runs in your browser · Nothing uploaded
- Instruction diffCompare two prompt or instruction versions line by line.Runs in your browser · Nothing uploaded
- System prompt builderAssemble a system prompt from role, rules, and output format sections.Runs in your browser · Nothing uploaded
- Prompt formatterClean up whitespace and wrap prompt sections in XML or Markdown.Runs in your browser · Nothing uploaded
- CLAUDE.md generatorDraft a CLAUDE.md project file for coding agents from a short form.Runs in your browser · Nothing uploaded
- AI rules generatorWrite rules files for Cursor, Copilot, and other assistants from one set of rules.Runs in your browser · Nothing uploaded
- MCP config builderBuild an MCP servers JSON config for a client without hand-editing.Runs in your browser · Nothing uploaded
- JSON schema builderDefine fields in a list and get a JSON Schema for structured model output.Runs in your browser · Nothing uploaded
- Skill builderCreate a SKILL.md with front matter and instructions for an agent skill.Runs in your browser · Nothing uploaded
- Prompt librarySave and organize prompts in this browser. Nothing leaves the device.Runs in your browser · Nothing uploaded
About AI Developer
These are the small chores around working with language models: how big is this prompt, what will it cost, what should the CLAUDE.md say. None of them call an API. Your prompts often contain internal code or customer text, which is a good reason not to paste them into yet another website.
Counts and costs are estimates. Tokenizers differ by vendor and prices change, so use the numbers to plan and leave a margin. Generators write text files; they don't install anything or contact any service.
Nothing on this page is saved between visits. The model tables and the generator forms start from defaults every time, so copy or download what you want to keep.
Tools in this category
Context window checker
Context window checker counts your text with the GPT-4o tokenizer (or takes a token count you type) and shows how full each model's window is. There is no output-reserve field, and the window sizes are editable starting values that can lag behind vendors.
LLM cost calculator
LLM cost calculator multiplies input and output token counts you enter by per-million prices in an editable model table. It does not count tokens, and it does not model caching or batch discounts.
Instruction diff
Instruction diff compares two prompt versions word by word, using the same engine as the text diff. Reformatting shows as a change, and there are no token counts.
System prompt builder
System prompt builder assembles role, context, task, rules, output-format, and example sections into one prompt, in Markdown sections or XML tags. It structures your text and does not call a model.
Prompt formatter
Prompt formatter trims stray whitespace and can wrap sections, found by short "Label:" lines or # headings, in XML tags or Markdown headings. It never rewrites your wording.
CLAUDE.md generator
CLAUDE.md generator drafts a project file for coding agents from a short form. It is a starting point; edit it with your real commands.
AI rules generator
AI rules generator turns one list of rules into five files: a Cursor rule, Copilot instructions, AGENTS.md, CLAUDE.md, and .windsurfrules. It only writes text.
MCP config builder
MCP config builder writes the mcpServers JSON a client reads, for command servers (arguments and environment) or HTTP URLs. It does not start or test the servers.
JSON schema builder
JSON schema builder defines flat fields in a list and writes a schema for structured output or tool calls, plain or in OpenAI strict form. Nested objects are not supported, and providers accept different subsets of JSON Schema.
Skill builder
Skill builder writes a SKILL.md with front matter and instructions. You still create the folder and any scripts.
Prompt library
Prompt library is fifteen built-in prompts for coding, writing, analysis, and agents, with search and a Copy button. You cannot add, edit, or save prompts here.
Common questions
Do these tools send my prompts anywhere?
No. Counting, formatting, and generating all run in the tab. There is no call to a model provider and no logging of what you paste.
How accurate is the token count?
Exact for OpenAI models when you pick their encoding. For Claude, Gemini, Llama and others it's an estimate, because those tokenizers aren't bundled here. Treat the figure as close, not exact, and keep headroom.
Are the model prices and context sizes current?
They were correct when the site was built. Vendors change both often, so check the provider's page for anything you're budgeting against.
Why these tools don't upload your files
There is no server to upload to. Each page is a static file, and the work happens in your browser. Your file is read from disk into memory, processed, and handed back as a download. Once the page has loaded you can disconnect from the network and it keeps working. The longer explanation