#compression

Claude Savings

Decompute

Cut your Claude Code token bill. A local gateway compresses context before each request — same model, same answers, fewer tokens. Bring your own key; runs entirely on your machine.

TokenSaver

maazizit

Compress AI shell output automatically — no TokViz install needed. Bundles TokViz, detects Cursor/Copilot/Gemini-Antigravity, installs hooks silently. Dashboard shows tokens saved. Unlike TokGuess (usage tracking), this actually reduces context sent to your agent.

Less Tokens for VS Code

shaminchokshi

A prompt compressor for Cursor / Copilot: compose, pick techniques, attach files (OCR / reduce), compress, and insert — fewer context tokens, same meaning.

吸T神功（XiT）

XiT

压缩高噪音终端输出，在 VS Code 状态栏和 Dashboard 中展示 AI 编程工作流的 Token 节省效果。

View, inspect, and convert .kore binary files directly in VS Code with table display, database import, and multi-format export

H47 Token Optimizer

wcalmels

Compress LLM prompts locally — no API calls. Reduce tokens on selected text for Claude, GPT, and Cursor.

LatentGate — Local-First AI Compression

KathanModh259

Compress images, text, conversations, and documents locally via Ollama before sending to cloud LLMs. Save ~80% on token costs.

ACIL — AI Credit Intelligence Layer

imKrisK

Pre-execution LLM cost governor for VS Code. Predicts cost before tokens are spent, compresses prompts, routes to cheaper models, enforces budgets, and learns your session patterns. Patent Pending.

Compress your prompts. Cut your token bill. Say less. Get more.