RTK (Rust Token Killer)

RTK (Rust Token Killer) is a powerful command-line tool designed to drastically cut down on token usage when interacting with Large Language Models (LLMs) like Claude Code or GitHub Copilot. It acts as a transparent proxy, intercepting your shell commands and intelligently compressing their output before it ever reaches your AI assistant. This means you can run commands like `git status`, `ls`, or `cargo test`, and RTK will automatically filter out verbose, irrelevant information, leaving only the crucial details for your LLM to process. The result is a significant reduction in token consumption, often ranging from 60% to 90%, which translates directly into lower API costs and faster LLM responses. RTK works by employing four core strategies: smart filtering, grouping, truncation, and deduplication. Smart filtering removes common noise like comments, excessive whitespace, and boilerplate text. Grouping aggregates similar items, such as files by directory or errors by type, to provide a concise overview. Truncation ensures that only the most relevant context is kept, cutting out redundancy. Finally, deduplication collapses repeated log lines, replacing them with a count, further reducing the output size. This multi-pronged approach ensures that the LLM receives only the essential information it needs to understand your project state or command output. Installation is straightforward, with options for Homebrew, a quick install script for Linux/macOS, Cargo, or pre-built binaries for various operating systems, including Windows. Once installed, users can initialize RTK for their specific AI tool (e.g., Claude Code, Gemini CLI, Cursor) using a simple `rtk init` command. This setup creates a transparent auto-rewrite hook in your shell, so commands like `git status` are automatically transformed into `rtk git status` without any manual intervention. Your LLM sees the compressed output, while you continue to use your familiar commands. RTK is ideal for developers, data scientists, and anyone who frequently uses LLMs for coding assistance, debugging, or general command-line tasks. It's particularly useful in environments where token costs are a concern or where large command outputs can quickly fill up an LLM's context window. Key features include token-optimized file and directory listings, compact Git command outputs (status, log, diff), streamlined GitHub CLI interactions, and highly compressed test runner results for tools like Jest, Pytest, and Cargo Test. It also supports build and lint tools, package managers, AWS CLI commands, and container tools like Docker and kubectl. Beyond compression, RTK provides token savings analytics, allowing users to track their savings and discover further opportunities for optimization.