Reduce agent token burn on repo-scale coding commands with the rtk CLI proxy
<p>Use rtk when an agent keeps wasting context on noisy shell output from large repos and you need smaller command responses before the model sees them.</p>
<p>Use rtk when an agent keeps wasting context on noisy shell output from large repos and you need smaller command responses before the model sees them.</p>
npx skills add agentskillexchange/skills --skill reduce-agent-token-burn-on-repo-scale-coding-commands-with-the-rtk-cli-proxy
rtk gives an agent a narrow shell-output compression workflow. It sits in front of commands like git status, git diff, grep, cat, and test runners, then filters and condenses the output before it reaches the model context. That makes it useful in repo-scale coding sessions where shell output, not reasoning, is what burns tokens.
The boundary is clear enough to be skill-shaped. Invoke it when the task is shrinking noisy command output for an existing coding-agent workflow, not when the user wants a new coding agent, an IDE extension, or a general terminal replacement. The job-to-be-done is context-saving shell mediation for large-codebase work.