rtk-token-saver

You can compress large shell and git command outputs to reduce token usage while preserving raw-output fallbacks. This approach optimizes AI agent performance during repository inspection, diffs, and log processing.

The best way to optimize AI agent performance for large repository inspection is to decrease the tokens required for commands like search, file-reading, test, build, and lint. This output compression significantly cuts token processing needs.

Yes, you need RTK installed and configured to compress command output for token savings. The Skill requires RTK to be set up in your environment before you can activate it for shell commands to reduce token consumption.

You can compress package manager and container logs to reduce token usage without losing data because the compression mechanism preserves raw-output fallbacks. This allows efficient processing of extensive build and lint outputs.

Processing extensive git history and diffs consumes many tokens because commands that produce large outputs inherently require more tokens for AI agents to parse. Compressing this output directly decreases the token consumption needed for processing.