Merge claude/practical-einstein-db4d4l into master

.github/workflows/server-shell-setup.yml

@@ -0,0 +1,40 @@
+# server-shell-setup is plain bash with no package.json, so it is outside the
+# turbo test matrix. This job runs its Vitest suites on a stock Ubuntu runner:
+# first the dry-run tests, then a real `--components all` install followed by
+# checks from inside bash, fish and nushell that everything landed.
+
+name: server-shell-setup
+
+on:
+  pull_request:
+    paths:
+      - "packages/server-shell-setup/**"
+      - ".github/workflows/server-shell-setup.yml"
+  push:
+    branches: [master]
+    paths:
+      - "packages/server-shell-setup/**"
+      - ".github/workflows/server-shell-setup.yml"
+  workflow_dispatch:
+
+jobs:
+  install-shell:
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+    defaults:
+      run:
+        working-directory: packages/server-shell-setup
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 22
+
+      - name: Unit tests (dry run, no system changes)
+        run: npx --yes vitest@3 run test/install-shell.test.js
+
+      - name: End-to-end install and verify every shell
+        env:
+          SERVER_SHELL_SETUP_E2E: "1"
+        run: npx --yes vitest@3 run test/install-shell.e2e.test.js

packages/about-system-info/README.md

@@ -202,7 +202,9 @@ async function getBasicInfo() {
 | `isp`         | Internet service provider  | `👮 Verizon Business`        |
 | `os`          | Operating system           | `⚡ Ubuntu 22.04`            |
 | `cpu`         | CPU information            | `📈 Intel Core i7-8700K`     |
+| `bench`       | CPU Geekbench 6 multi-core, rank, top percentile | `💪 7k #612 top 62%` |
 | `gpu`         | Graphics card              | `🎮 NVIDIA GeForce RTX 3080` |
+| `gpu_bench`   | GPU Geekbench 6 OpenCL, rank, top percentile | `🏎️ 175k #140 top 14%` |
 | `disk_used`   | Disk usage percentage      | `📁 75%`                     |
 | `disk_size`   | Disk size per real disk    | `💽 256/512GB`               |
 | `ram_used`    | Memory usage               | `💾 8/16GB`                  |

packages/about-system-info/src/about-system-cli.ts

@@ -478,7 +478,7 @@ Available display blocks:
   Basic: user, hostname, uptime, shell, os, kernel, device
   Resources: disk_used, ram_used, memory_available, swap_used, top_process
   Network: ip, iplocal, city, domain, isp, network_interfaces
-  Hardware: cpu, gpu, temperature, battery, screen_resolution
+  Hardware: cpu, bench, gpu, gpu_bench, temperature, battery, screen_resolution
   System: load_average, users_logged_in, mount_points, services_running
   Tools: pacman, ports, containers
 `);

packages/about-system-info/src/info/hardware.test.ts

@@ -0,0 +1,28 @@
+import { describe, expect, it } from "vitest";
+import { formatBench, topPercent } from "./hardware";
+
+describe("topPercent", () => {
+  it("rounds up so the fastest entry reads top 1%", () => {
+    expect(topPercent(1, 1000)).toBe(1);
+    expect(topPercent(68, 1000)).toBe(7);
+    expect(topPercent(50, 1000)).toBe(5);
+  });
+
+  it("stays within 1..100", () => {
+    expect(topPercent(1000, 1000)).toBe(100);
+    expect(topPercent(1200, 1000)).toBe(100);
+    expect(topPercent(5, 0)).toBe(100);
+  });
+});
+
+describe("formatBench", () => {
+  it("rounds the score to whole thousands and appends the percentile", () => {
+    expect(formatBench(37967, 1, 1000)).toBe("38k #1 top 1%");
+    expect(formatBench(20512, 68, 1000)).toBe("21k #68 top 7%");
+    expect(formatBench(386779, 5, 1026)).toBe("387k #5 top 1%");
+  });
+
+  it("never prints a decimal in the score", () => {
+    expect(formatBench(19249, 91, 1000)).not.toMatch(/\d\.\d/);
+  });
+});

packages/about-system-info/src/info/hardware.ts

@@ -11,6 +11,30 @@ import { execCommand } from "../utils/command";
 import { getCachedValue, setCachedValue } from "../cache/cache";
 import Fuse from "fuse.js";
 
+/**
+ * Formats a Geekbench score in whole thousands with its rank and how far up
+ * the table that rank sits.
+ * @param score - Raw Geekbench score
+ * @param rank - Rank in the benchmark table (1 = fastest)
+ * @param total - Number of entries in the benchmark table
+ * @returns Score, rank and top percentile
+ * @example formatBench(20512, 68, 1500) // "21k #68 top 7%"
+ */
+export function formatBench(score: number, rank: number, total: number): string {
+  return `${Math.round(score / 1000)}k #${rank} top ${topPercent(rank, total)}%`;
+}
+
+/**
+ * Percentile bucket a rank falls in, rounded up so rank 1 reads "top 1%"
+ * @param rank - Rank in the benchmark table (1 = fastest)
+ * @param total - Number of entries in the benchmark table
+ * @returns Whole percentage between 1 and 100
+ */
+export function topPercent(rank: number, total: number): number {
+  if (!total || total < 1) return 100;
+  return Math.min(100, Math.max(1, Math.ceil((rank / total) * 100)));
+}
+
 /**
  * Gets CPU model name and specifications
  * Uses platform-specific commands (wmic/lscpu/cpuinfo)
@@ -222,7 +246,7 @@ export function screen_resolution(): string {
  * Uses fuzzy matching to find similar CPU models
  * @param context - Info context with cache
  * @returns Benchmark info with score and rank, or empty string
- * @example "37.9k #1", "20.5k #68"
+ * @example "38k #1 top 1%", "21k #68 top 7%"
  */
 export function bench(context: InfoContext): string {
   const cached = getCachedValue(context.cache, "bench");
@@ -274,9 +298,7 @@ export function bench(context: InfoContext): string {
     const results = fuse.search(cpuName);
     if (results.length > 0) {
       const match = results[0].item;
-      const score = Math.round(match.score / 100) / 10;
-      const rank = match.rank;
-      const result = `${score}k #${rank}`;
+      const result = formatBench(match.score, match.rank, scores.length);
       setCachedValue(context.cache, "bench", result);
       return result;
     }
@@ -293,7 +315,7 @@ export function bench(context: InfoContext): string {
  * Includes score, rank, and architecture type (ARM/Intel/AMD)
  * @param context - Info context with cache
  * @returns Detailed benchmark info with architecture, or empty string
- * @example "Geekbench 6: 37.9k (Rank #1) - ARM", "Geekbench 6: 20.5k (Rank #68) - Intel"
+ * @example "Geekbench 6: 38k (Rank #1, top 1%) - ARM", "Geekbench 6: 21k (Rank #68, top 7%) - Intel"
  */
 export function cpu_bench_info(context: InfoContext): string {
   const cached = getCachedValue(context.cache, "cpu_bench_info");
@@ -349,8 +371,9 @@ export function cpu_bench_info(context: InfoContext): string {
     const results = fuse.search(cpuName);
     if (results.length > 0) {
       const match = results[0].item;
-      const score = Math.round(match.score / 100) / 10;
+      const score = Math.round(match.score / 1000);
       const rank = match.rank;
+      const top = topPercent(rank, scores.length);
 
       // Detect architecture type
       let arch = "Unknown";
@@ -364,7 +387,7 @@ export function cpu_bench_info(context: InfoContext): string {
         arch = "ARM";
       }
 
-      const result = `Geekbench 6: ${score}k (Rank #${rank}) - ${arch}`;
+      const result = `Geekbench 6: ${score}k (Rank #${rank}, top ${top}%) - ${arch}`;
       setCachedValue(context.cache, "cpu_bench_info", result);
       return result;
     }
@@ -381,7 +404,7 @@ export function cpu_bench_info(context: InfoContext): string {
  * Uses fuzzy matching to find similar GPU models
  * @param context - Info context with cache
  * @returns Benchmark info with score and rank, or empty string
- * @example "37.9k #1", "20.5k #68"
+ * @example "38k #1 top 1%", "21k #68 top 7%"
  */
 export function gpu_bench(context: InfoContext): string {
   const cached = getCachedValue(context.cache, "gpu_bench");
@@ -433,9 +456,7 @@ export function gpu_bench(context: InfoContext): string {
     const results = fuse.search(gpuName);
     if (results.length > 0) {
       const match = results[0].item;
-      const score = Math.round(match.score / 100) / 10;
-      const rank = match.rank;
-      const result = `${score}k #${rank}`;
+      const result = formatBench(match.score, match.rank, scores.length);
       setCachedValue(context.cache, "gpu_bench", result);
       return result;
     }
@@ -452,7 +473,7 @@ export function gpu_bench(context: InfoContext): string {
  * Includes score, rank, and vendor detection
  * @param context - Info context with cache
  * @returns Detailed benchmark info with vendor, or empty string
- * @example "Geekbench 6: 37.9k (Rank #1) - NVIDIA", "Geekbench 6: 20.5k (Rank #68) - AMD"
+ * @example "Geekbench 6: 38k (Rank #1, top 1%) - NVIDIA", "Geekbench 6: 21k (Rank #68, top 7%) - AMD"
  */
 export function gpu_bench_info(context: InfoContext): string {
   const cached = getCachedValue(context.cache, "gpu_bench_info");
@@ -508,8 +529,9 @@ export function gpu_bench_info(context: InfoContext): string {
     const results = fuse.search(gpuName);
     if (results.length > 0) {
       const match = results[0].item;
-      const score = Math.round(match.score / 100) / 10;
+      const score = Math.round(match.score / 1000);
       const rank = match.rank;
+      const top = topPercent(rank, scores.length);
 
       // Detect vendor from GPU name
       let vendor = "Unknown";
@@ -532,7 +554,7 @@ export function gpu_bench_info(context: InfoContext): string {
         vendor = "Intel";
       }
 
-      const result = `Geekbench 6: ${score}k (Rank #${rank}) - ${vendor}`;
+      const result = `Geekbench 6: ${score}k (Rank #${rank}, top ${top}%) - ${vendor}`;
       setCachedValue(context.cache, "gpu_bench_info", result);
       return result;
     }

packages/about-system-info/src/info/settings.ts

@@ -59,7 +59,17 @@ export const backgrounds = {
 // Default settings
 export const DEFAULT_SETTINGS = {
   display_order: [
-    ["user", "hostname", "os", "device", "kernel", "cpu", "gpu", "bench"],
+    [
+      "user",
+      "hostname",
+      "os",
+      "device",
+      "kernel",
+      "cpu",
+      "bench",
+      "gpu",
+      "gpu_bench",
+    ],
     [
       "disk_used",
       "disk_size",
@@ -90,6 +100,7 @@ export const DEFAULT_SETTINGS = {
     cpu: "orange",
     gpu: "yellow",
     bench: "red",
+    gpu_bench: "red",
     device: "yellow",
     kernel: "green",
     shell: "orange",
@@ -119,6 +130,7 @@ export const DEFAULT_SETTINGS = {
     cpu: "📈 ",
     gpu: "🎮 ",
     bench: "💪 ",
+    gpu_bench: "🏎️ ",
     device: "💻 ",
     kernel: "🔧 ",
     shell: "🐚 ",
@@ -154,6 +166,7 @@ export const DEFAULT_SETTINGS = {
     cpu: "CPU",
     gpu: "GPU",
     bench: "Bench",
+    gpu_bench: "GPU Bench",
     device: "Device",
     kernel: "Kernel",
     shell: "Shell",

packages/about-system-info/src/systeminfo-types.d.ts

@@ -129,8 +129,8 @@ export interface SystemInfo {
    * Empty string if CPU not found or benchmark data unavailable
    *
    * @example
-   * - "37.9k #1" (top performing CPU)
-   * - "20.5k #68" (mid-range CPU)
+   * - "38k #1 top 1%" (top performing CPU)
+   * - "21k #68 top 7%" (mid-range CPU)
    * - "" (CPU not in benchmark database)
    */
   bench: string;
@@ -142,9 +142,9 @@ export interface SystemInfo {
    * Empty string if CPU not found or benchmark data unavailable
    *
    * @example
-   * - "Geekbench 6: 37.9k (Rank #1) - ARM" (Apple M4 Ultra)
-   * - "Geekbench 6: 20.5k (Rank #68) - Intel" (Intel Core i9)
-   * - "Geekbench 6: 19.2k (Rank #91) - AMD" (AMD Ryzen 9 7950X)
+   * - "Geekbench 6: 38k (Rank #1, top 1%) - ARM" (Apple M4 Ultra)
+   * - "Geekbench 6: 21k (Rank #68, top 7%) - Intel" (Intel Core i9)
+   * - "Geekbench 6: 19k (Rank #91, top 10%) - AMD" (AMD Ryzen 9 7950X)
    * - "" (CPU not in benchmark database)
    */
   cpu_bench_info: string;
@@ -156,8 +156,8 @@ export interface SystemInfo {
    * Empty string if GPU not found or benchmark data unavailable
    *
    * @example
-   * - "37.9k #1" (top performing GPU)
-   * - "20.5k #68" (mid-range GPU)
+   * - "38k #1 top 1%" (top performing GPU)
+   * - "21k #68 top 7%" (mid-range GPU)
    * - "" (GPU not in benchmark database)
    */
   gpu_bench: string;
@@ -169,8 +169,8 @@ export interface SystemInfo {
    * Empty string if GPU not found or benchmark data unavailable
    *
    * @example
-   * - "Geekbench 6: 37.9k (Rank #1) - NVIDIA" (NVIDIA GPU)
-   * - "Geekbench 6: 20.5k (Rank #68) - AMD" (AMD GPU)
+   * - "Geekbench 6: 38k (Rank #1, top 1%) - NVIDIA" (NVIDIA GPU)
+   * - "Geekbench 6: 21k (Rank #68, top 7%) - AMD" (AMD GPU)
    * - "" (GPU not in benchmark database)
    */
   gpu_bench_info: string;

packages/server-shell-setup/README.md

@@ -43,12 +43,29 @@ One-command setup for a modern dev environment: `fish`, `nvim`, `nushell`, `bun`
 
 ## Install
 
-On a fresh server, you may first need to set passwords:
+On a fresh server, the installer checks whether root has a password before it
+installs anything. If root has none, it prompts you to set one on a terminal;
+in an unattended run it sets root to the user password you pass in. The root
+and user password can be the same:
 ```bash
-sudo passwd        # set root password
-sudo passwd $USER  # set user password
+# One password for both root and your user (env var keeps it out of `ps`)
+SETUP_PASSWORD='...' bash install-shell.sh --components all --yes
+
+# Or separately
+bash install-shell.sh --root-password '...' --user-password '...' --components all --yes
+
+# Only set passwords, install nothing
+SETUP_PASSWORD='...' bash install-shell.sh --yes
 ```
 
+| Option | Env var | Effect |
+|--------|---------|--------|
+| `--root-password PW` | `SETUP_ROOT_PASSWORD` | Always sets root's password |
+| `--user-password PW` | `SETUP_USER_PASSWORD` | Sets your user's password, and root's when root has none |
+| `--password PW` | `SETUP_PASSWORD` | Same value for both |
+
+An existing root password is only replaced when `--root-password`/`--password` is given.
+
 **Interactive menu** — pick what to install:
 ```bash
 wget -qO- tinyurl.com/shellsetup | bash
@@ -518,7 +535,7 @@ It finishes by syncing and dropping the kernel page cache. The Docker prune dele
 
 | Symptom | Fix |
 |---------|-----|
-| Installer aborts immediately on a fresh server | Set passwords first (`sudo passwd`, `sudo passwd $USER`); many images ship with none, which makes `sudo` fail confusingly |
+| Installer aborts immediately on a fresh server | Many images ship with no passwords, which makes `sudo` fail confusingly. Re-run with `SETUP_PASSWORD=... ` (or `--password`) so the installer sets them first |
 | Piped install shows the menu instead of running unattended | You dropped `-s --` — it's `bash -s -- all`, not `bash all` |
 | Shell is still bash after login | `chsh` may have failed silently; run `chsh -s $(which fish)` and open a new session |
 | `Invalid component: <name>` | Check the name against the valid list above; the script exits rather than guessing |