ojuschugh1

ojuschugh1 / sqz

Public

Compress LLM context to save tokens and reduce costs

ojuschugh1.github.iosqz agentic-ai ai api cli context
104
5
100% credibility
Found Apr 15, 2026 at 16 stars 5x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

sqz compresses text, code, and command outputs sent to AI models to drastically reduce token usage and costs, working seamlessly via shell integration, browser extensions for ChatGPT/Claude/Gemini/Grok/Perplexity, and IDE plugins for VS Code and JetBrains.

How It Works

1
🔍 Discover sqz

You hear about sqz, a smart helper that shrinks text before sending it to AI chats like ChatGPT or Claude, saving you money on every conversation.

2
📥 Get sqz

Download and add sqz to your computer with a simple command, like grabbing a helpful app.

3
⚙️ Set it up

Run a quick setup that adds sqz to your coding tools, browser, and command line so it works everywhere automatically.

4
✨ Paste and squeeze

Paste code or output into your AI chat—sqz shows a preview of the shorter version and how many words it saved.

5
✅ Send smarter

Click to use the squeezed version; your AI gets the key info faster without wasting space.

📊 Track your savings

Check your daily chart to see tokens and dollars saved, feeling great about smarter, cheaper AI use.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 104 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is sqz?

sqz compresses LLM context like CLI outputs, file contents, and pasted prompts to slash token usage and costs. Built in Rust as a single binary, it hooks into shells for git/cargo/npm commands, runs browser extensions on ChatGPT/Claude/Gemini/Grok/Perplexity, and adds IDE plugins for VS Code/JetBrains with token budget tracking. Users run `sqz init` once, then get deduped file refs (13 tokens on repeats) and commands like `sqz gain` for savings charts.

Why is it gaining traction?

Unlike llmlingua or basic prompt compressors, sqz handles full workflows—compress llm context from 150+ CLI patterns, json for llm APIs, and repeated reads—delivering 58-86% savings on logs/diffs/files without workflow changes. Browser extensions auto-squeeze pastes client-side (zero net), MCP server picks tools dynamically, and USD tracking shows real ROI. Zero telemetry and offline mode appeal to privacy-focused devs.

Who should use this?

AI agent users on Cursor/Claude Code/Aider hitting token walls during code reviews or debug sessions. Devs piping git diffs/cargo test outputs to LLMs, or pasting code into web UIs. Teams optimizing multi-tool chains like docker/kubectl with prompts.

Verdict

Early but solid for compress llm prompt needs—586 tests and clear docs offset 12 stars/1.0% credibility; trial it if tokens hurt, but verify in your stack first.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.