rtk-exec

You can compress CLI output to save LLM tokens by executing shell commands through the RTK proxy, which filters, groups, and truncates verbose diagnostic data by 60-90% before it reaches the AI.

Yes, the RTK proxy works with kubectl, pytest, and over 100 common CLI tools including git and cargo. It wraps these standard shell commands to generate token-optimized, compressed output for broad compatibility.

The best way to fit large test suite results into constrained context windows is using a token-optimized proxy to compress the verbose diagnostic data, reducing token consumption by 60-90% for deeper AI analysis without hitting limits.

Yes, you need the rtk binary installed and accessible in your system path to run shell commands with token compression. This prerequisite allows the skill to proxy standard commands through the RTK engine.

You should not use a token compression proxy when you require the complete, unmodified CLI output for strict auditing, as the RTK engine actively filters, groups, and truncates verbose data to optimize tokens.

Verbose CLI output exhausts context windows quickly because standard tools generate massive amounts of text from system logs and test results, increasing token consumption and costs before the AI can process the data.