Count tokens for GPT, Claude, and Gemini using official tokenizers.
npx skills add https://github.com/ShipFail/promptware --skill stoptimizer
Please help me install this Agent Skill. Skill: StopTimizer Source: https://github.com/ShipFail/promptware/tree/main/os/skills/stoptimizer Command: npx skills add https://github.com/ShipFail/promptware --skill stoptimizer
StopTimizer counts tokens for GPT, Claude, and Gemini using official tokenizers as source of truth.
Use stoptimizer.ts to count tokens for your input text: deno run --allow-net <url>/stoptimizer.ts "hello world" By default, it returns space-delimited counts for all supported models. Use --json for a JSON object, or --model to query a single model.
High-intent search queries and answers about installing and using this skill.
To count tokens accurately across GPT and Claude, StopTimizer uses official external tokenizers like gpt-tokenizer and the Anthropic tokenizer as its source of truth for precise prompt design and budgeting.
You can get JSON output for AI workflow automation by running the Deno script with the --json flag, which returns a structured JSON object of token counts instead of the default space-delimited format.
Yes, StopTimizer supports Deno for counting LLM tokens. You can execute the script directly using the deno run command with network permissions to fetch required external tokenizer dependencies.
The best way to validate STOP protocol token limits is by querying all supported models simultaneously, returning space-delimited counts to quickly verify if your prompt design fits within strict AI model budgets.
Yes, you can check token counts for a single specific model by using the --model flag, which restricts the tokenization query to only the AI model you specify instead of returning counts for all.
xberg-ioSplit text into chunks by size, overlap, and chunker type.
devendrapratapsinghOptimize AI and LLM operations for token usage, cost, and inference.
Prathmesh2000Analyze and estimate token usage for LLM API calls.
ByronWilliamsCPATrack Claude Code token usage across local JSONL transcripts.
