Nyquest – LLM Prompt Compression

Compress LLM prompts by 15-25% before they leave your browser. 281 rules, zero server required.

Nyquest compresses your LLM prompts before they leave your browser — saving 15-25% on tokens with zero impact on response quality. HOW IT WORKS When you send a message on Claude, ChatGPT, Gemini, or Grok, Nyquest intercepts the outbound request and applies 281 compression rules that remove filler words, verbose phrases, and redundant formatting. The compressed prompt is semantically equivalent — the AI gives the same answer, you just pay fewer tokens for it. SUPPORTED PROVIDERS - Claude (claude.ai) - ChatGPT (chatgpt.com) - Gemini (gemini.google.com) - Grok (grok.com) WHAT GETS COMPRESSED - Filler phrases: "it is important to note that" → removed - Verbose expressions: "due to the fact that" → "because" - Imperative bloat: "you should always make sure to" → removed - Conversational padding: "I hope this helps" → removed - Contractions: "do not" → "don't" - Markdown noise: excessive bold, deep heading levels - And 275+ more rules across 18 categories WHAT IT DOESN'T TOUCH - Code blocks and technical content - Tool calls, image references, file attachments - Message IDs, conversation state, model parameters - Any content under 20 characters FEATURES - Real-time HUD overlay showing compression stats - Popup dashboard with session statistics and live feed - Adjustable compression level (gentle → aggressive) - Test panel to preview compression on any text - Works instantly — no account, no server, no signup 100% LOCAL All compression runs in your browser. No data is sent to Nyquest servers. No analytics. No telemetry. Your prompts stay between you and your AI provider. OPEN & TRANSPARENT The compression rules are readable JavaScript — no obfuscation. You can inspect exactly what the extension does at any time.