ojuschugh1Developer tools
sqz
A Rust-based CLI and MCP tool that compresses command output and deduplicates context to cut LLM token costs.
- CLI
- IDE plugin
- Browser extension
- Library/framework
- Windows
- macOS
- Linux
- Browser
- Self-hostable
- Runs locally
- Docker supported
- Popularity
- 624 Stars
- GitHub stars
- Recent activity
- 9/15/2026
- Updated in the last 30 days
- License
- NOASSERTION
- Review the terms yourself
Why it matters
We look beyond stars: what problem it solves, whether it creates real utility, and what makes its approach worth noticing.
Problem
AI coding agents repeatedly read identical files and console logs across context turns, leading to rapid token consumption and higher API costs.
Practical value
Provides a single Rust binary and MCP server that intercepts tool outputs, substituting repeated file reads with 13-token reference keys.
Innovation / differentiation
Combines command-level structured compression with session-level deduplication that replaces repeated content at the tool hook layer.
Leverage potential
Built on a Rust engine with automated hook configuration for various AI editors and agents, enabling low-friction distribution.
Why now
The rise of CLI agents like Claude Code and OpenCode has intensified context growth from tool outputs, making compression timely.
Community activity
In the last 90 days there were 8 new issues and 3 pull requests; the bounded issue/PR samples include 8 issue authors and 3 PR contributors, with 5 releases.
Maintainer responsiveness
The 8-issue window sample had a 100% close rate, and the 3-pull-request sample had a 100% merge rate. Maintainer-comment observations covered only 100% of that issue sample, so response rate and first-response speed are not reported.
Key highlights
- Single Rust binary that intercepts and compresses command outputs
- Session deduplication reducing repeated file reads to 13-token references
- Automatic hook setup for Claude Code, Cursor, Windsurf, Gemini, OpenCode, and Codex
Quick start
How it is installed, how hard it is, and where to start.
Where it runs
To be confirmed
Difficulty
To be confirmed
- 01Download and install sqz binary using the installation script.
- 02Initialize sqz hooks globally across all AI agent session configs on your machine.
- 03Limit hook configuration to specific agents using the --only flag.
- 04Run the gain analysis command to review total token savings and reduction percentages.
Best for
- Teams that want data on their own servers
- Developers who want to try it on their machine
- People who prefer Docker deploys
More about it
sqz is an open-source context compression tool written in Rust, designed to minimize LLM token consumption during AI agent workflows. During coding sessions, agents frequently re-read the same source files or generate repetitive command outputs; sqz intercepts these outputs before they reach the language model. Its key mechanism is session-level deduplication: when a file is read multiple times in one session, sqz sends full contents once and converts subsequent reads into 13-token reference markers. It also compresses test logs, JSON structures, and git diffs on a per-command basis. The tool integrates directly into popular agent environments including Claude Code, Cursor, Windsurf, OpenCode, and Codex.
Sources
Each field shows its status and source — expand to review.
15 · Expand
Sources
Each field shows its status and source — expand to review.
capability tags
VerifiedSource: admin_cms · cms editor · 8/30/2026
editor note
Verified{"en":"When running AI coding agents like Claude Code or Cursor, repeated file reads and verbose logs quickly exhaust your context window. sqz intercepts CLI outputs, compresses log streams, and replaces cached file contents with 13-token reference tags. It installs as a single Rust binary and merges into existing agent configurations.","zh":"在运行 Claude Code、Cursor 等 AI 编程助手时，大量日志与重复读取同一文件会快速消耗上下文额度。sqz 通过 Rust 单文件可执行程序拦截终端输出，对于重复读取的文件仅保留 13 个 Token 的引用标识。配置过程直接且支持自动注入配置文件，能有效控制 AI 开发中的 Token 消耗。"}
Source: admin_cms · cms editor · 8/30/2026
how to use
Verified{"steps":[{"en":"Download and install sqz binary using the installation script.","zh":"使用一键安装脚本下载并安装 sqz 可执行文件。"},{"en":"Initialize sqz hooks globally across all AI agent session configs on your machine.","zh":"在全局范围初始化 sqz 钩子，将其应用于本机的 AI Agent 配置文件中。"},{"en":"Limit hook configuration to specific agents using the --only flag.","zh":"如果仅需要向特定 Agent（如 OpenCode 或 Codex）添加钩子，可以使用限定参数。"},{"en":"Run the gain analysis command to review total token savings and reduction percentages.","zh":"运行收益统计命令，查看最近 7 天内节省的 Token 数量与压缩比例。"}],"installAt":"unknown","difficulty":"unknown"}
Source: admin_cms · cms editor · 8/30/2026
intro
Verified{"en":"sqz is an open-source context compression tool written in Rust, designed to minimize LLM token consumption during AI agent workflows. During coding sessions, agents frequently re-read the same source files or generate repetitive command outputs; sqz intercepts these outputs before they reach the language model. Its key mechanism is session-level deduplication: when a file is read multiple times in one session, sqz sends full contents once and converts subsequent reads into 13-token reference markers. It also compresses test logs, JSON structures, and git diffs on a per-command basis. The tool integrates directly into popular agent environments including Claude Code, Cursor, Windsurf, OpenCode, and Codex.","zh":"sqz 是一款基于 Rust 语言开发的 LLM 上下文压缩工具，专门处理 AI Agent 在执行代码任务时产生的多余 Token。在代码编写或调优过程中，AI Agent 往往会多次读取相同的文件或产生大量冗余日志，sqz 能够在命令行输出传入模型之前进行拦截与优化。 其核心功能是会话级的去重机制。当相同文件被多次读取时，sqz 仅在首次发送完整内容，后续读取替换为 13 个 Token 的轻量级引用。此外，sqz 还能针对 JSON 接口返回值、测试日志和 Git diff 进行内容压缩。用户可通过命令行将配置关联至 Claude Code、Cursor、Windsurf 等 Agent 工具。"}
Source: admin_cms · cms editor · 8/30/2026
Latest release
Verifiedv1.6.1
Source: GitHub API · latest_release=v1.6.1 · 9/19/2026
License
VerifiedNOASSERTION
Source: GitHub API · license.spdx_id=NOASSERTION · 9/19/2026
needs api key
VerifiedNo
Source: admin_cms · cms editor · 8/30/2026
One-liner
Verified{"en":"A Rust-based CLI and MCP tool that compresses command output and deduplicates context to cut LLM token costs.","zh":"基于 Rust 的 LLM 上下文压缩与去重工具，降低命令行与 AI Agent 的 Token 支出。"}
Source: admin_cms · cms editor · 8/30/2026
Platforms
Verifiedwindows, macos, linux, browser
Source: admin_cms · cms editor · 8/30/2026
Category hint
Inferred from materialsai-apps
Source: Project README · hint=ai-apps · 8/16/2026
product forms
Verifiedcli, ide_plugin, browser_extension, library_framework
Source: admin_cms · cms editor · 8/30/2026
role tags
VerifiedSource: admin_cms · cms editor · 8/30/2026
supports docker
VerifiedYes
Source: Repository file · dockerfile=true; compose=false · 9/19/2026
supports local
VerifiedYes
Source: admin_cms · cms editor · 8/30/2026
supports self host
VerifiedYes
Source: admin_cms · cms editor · 8/30/2026
Related projects
Other verified projects matched by category, capabilities, and intended roles.
claude-code
An agentic coding tool for the terminal that understands codebases, handles routine development tasks, explains code, and works with Git workflows.
spec-kit
An open-source CLI toolkit for Spec-Driven Development that guides AI coding agents through structured specification, planning, and code generation steps.
llama.cpp
A lightweight LLM inference engine implemented in C/C++, supporting various hardware backends and model quantization.
codex
OpenAI’s coding agent that runs locally, with terminal, IDE, and desktop-app entry points.
graphify
Convert codebases, docs, and configs into a queryable knowledge graph for AI coding assistants
supabase
An open-source Postgres development platform offering auto-generated APIs, authentication, and real-time database subscriptions.