@xynogen/pix-optimizer 1.1.27 → 1.1.29

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@xynogen/pix-optimizer",
-	"version": "1.1.27",
+	"version": "1.1.29",
 	"description": "Performance optimization suite for Pi Coding Agent - caveman mode + RTK tool rewriting + ponytail lazy-dev mode",
 	"type": "module",
 	"main": "src/index.ts",
@@ -9,7 +9,7 @@
 	},
 	"files": [
 		"src",
-		"!src/**/*.test.ts",
+		"!src/**/*.test.*",
 		"README.md",
 		"LICENSE"
 	],

package/src/rtk.ts

@@ -90,15 +90,6 @@ export function applyRtkRewrite(
 	return true;
 }
 
-const RTK_SYSTEM_PROMPT = `# RTK — token-optimized command wrapper
-
-Prefix shell commands with \`rtk\` (e.g. \`rtk git status\`). RTK compacts output for git, gh, cargo, npm/pnpm/yarn/bun, tsc, lint, vitest/jest/playwright, docker, kubectl, ls, grep, prisma — and passes anything else through unchanged, so it's always safe.
-
-Prefix EVERY segment in a chain, not just the first:
-\`rtk git add . && rtk git commit -m "msg" && rtk git push\`
-
-RTK also has filtering subcommands the auto-rewriter won't add — reach for these yourself when useful: \`rtk err <cmd>\` (errors only), \`rtk summary <cmd>\`, \`rtk log <file>\` (dedup), \`rtk json <file>\` (structure), \`rtk test <cmd>\` (failures only), \`rtk gain\` (savings stats).`;
-
 // Commands that should be prefixed with rtk
 const RTK_COMMANDS = new Set([
 	"git",
@@ -256,15 +247,15 @@ export function rtk(pi: ExtensionAPI, status: OptimizerStatus): OptimizerHandle
 		return rtkStatus;
 	};
 
-	// Inject RTK system prompt + detect sudo_run tool availability.
-	pi.on("before_agent_start", async (event) => {
-		// Detect sudo_run from the active tool list for this session.
+	// Detect sudo_run tool availability. No system-prompt injection — the
+	// tool_call rewrite hook adds the rtk prefix on its own, so we keep zero
+	// always-on context cost (ponytail: manual `rtk err`/`rtk summary`/`rtk
+	// test` subcommands are no longer advertised to the model; re-add a prompt
+	// here if you want those surfaced).
+	pi.on("before_agent_start", (event) => {
 		const tools: string[] = event.systemPromptOptions?.selectedTools ?? [];
 		hasSudoRunTool = tools.includes("sudo_run");
-
-		if (!enabled) return undefined;
-		const existing = event.systemPrompt ?? "";
-		return { systemPrompt: `${RTK_SYSTEM_PROMPT}\n\n${existing}` };
+		return undefined;
 	});
 
 	// Keep the status indicator in sync across the agent lifecycle. Probe