rtk-token-optimizer.json

{
	"schemaVersion": 1,
	"id": "rtk-token-optimizer",
	"enabledByDefault": false,
	"title": "RTK binary",
	"description": "Token-optimized command proxy for agent shell commands. Use it when workflow output is noisy and the operator wants shorter command logs without losing actionable failures.",
	"functional": [
		"Runs shell commands through the rtk binary.",
		"Filters repetitive terminal output before it enters the agent context.",
		"Reports token savings through rtk gain."
	],
	"source": {
		"type": "github",
		"githubRepo": "https://github.com/rtk-ai/rtk",
		"script": ""
	},
	"install": {
		"kind": "manual",
		"commands": [],
		"notes": [
			"Install rtk from the repository release or install instructions.",
			"Place the rtk binary on PATH before running devos onboard --check."
		]
	},
	"enable": {
		"kind": "path-binary",
		"config": {},
		"notes": [
			"devos onboard --check validates rtk --version.",
			"Agent instructions can opt in by prefixing shell commands with rtk."
		]
	},
	"checks": [
		{
			"title": "Verify RTK is available",
			"command": "rtk",
			"args": ["--version"],
			"expected": "exit code 0"
		},
		{
			"title": "Inspect token savings",
			"command": "rtk",
			"args": ["gain"],
			"expected": "prints token savings analytics"
		}
	],
	"tokenOptimization": {
		"strategy": "Reduce command-output tokens by proxying shell commands through rtk.",
		"savingsSignal": "Use rtk gain or rtk gain --history to compare filtered output savings.",
		"whenToUse": [
			"Workflow runs with verbose tests, builds, git output, or package-manager logs.",
			"Long-running agent sessions where terminal output dominates context."
		]
	},
	"maintainers": ["devos-cli"]
}