@kernel.chat/kbot 3.29.0 → 3.30.1

package/dist/cli.js

@@ -883,6 +883,69 @@ async function main() {
             process.stderr.write(formatMachineProfile(profile));
         }
     });
+    program
+        .command('hardware')
+        .description('Detect your hardware tier and get personalized model recommendations for local AI')
+        .action(async () => {
+        const chalk = (await import('chalk')).default;
+        const { detectHardwareTier, getMultiModelConfig, QUANT_OPTIONS, DEFAULT_MODELS } = await import('./inference.js');
+        const hw = detectHardwareTier();
+        const mm = getMultiModelConfig();
+        console.log();
+        console.log(`  ${chalk.bold('kbot hardware')} — your local AI capability`);
+        console.log();
+        const tierColors = {
+            basic: chalk.dim, standard: chalk.white, pro: chalk.cyan, ultra: chalk.hex('#A78BFA'), datacenter: chalk.hex('#FFD700'),
+        };
+        const colorFn = tierColors[hw.tier] || chalk.white;
+        console.log(`  ${chalk.bold('Tier')}: ${colorFn(hw.tier.toUpperCase())}`);
+        console.log(`  ${chalk.dim(hw.description)}`);
+        console.log(`  ${chalk.bold('Max model')}: ${hw.maxModelParams}`);
+        console.log();
+        // Model catalog by tier
+        console.log(`  ${chalk.bold('Model Catalog')} ${chalk.dim(`(${Object.keys(DEFAULT_MODELS).length} models)`)}`);
+        console.log(`  ${chalk.dim('─'.repeat(50))}`);
+        const tiers = [
+            { label: 'Light (2-4 GB)', filter: tags => tags.includes('lightweight') || tags.includes('fast') },
+            { label: 'Standard (4-6 GB)', filter: tags => !tags.includes('lightweight') && !tags.includes('fast') && !tags.includes('large') && !tags.includes('frontier') && !tags.includes('ultra') && !tags.includes('legacy') },
+            { label: 'Heavy (8-16 GB)', filter: tags => tags.includes('large') },
+            { label: 'Frontier (32-64 GB)', filter: tags => tags.includes('frontier') },
+            { label: 'Ultra (100+ GB)', filter: tags => tags.includes('ultra') },
+        ];
+        for (const tier of tiers) {
+            const models = Object.entries(DEFAULT_MODELS).filter(([, m]) => tier.filter(m.tags));
+            if (models.length === 0)
+                continue;
+            console.log(`  ${chalk.bold(tier.label)}`);
+            for (const [name, model] of models) {
+                const fits = parseFloat(model.size.replace(/[^0-9.]/g, '')) <= mm.totalRAM * 0.6;
+                const icon = fits ? chalk.green('✓') : chalk.red('✗');
+                const rec = model.tags.includes('recommended') ? chalk.yellow(' ★') : '';
+                console.log(`  ${icon} ${name.padEnd(22)} ${chalk.dim(model.size.padEnd(10))} ${model.description.slice(0, 55)}${rec}`);
+            }
+            console.log();
+        }
+        // Multi-model config
+        if (mm.canMultiModel) {
+            console.log(`  ${chalk.bold('Multi-Model Setup')} ${chalk.dim('(run two models simultaneously)')}`);
+            console.log(`  ${chalk.dim('─'.repeat(50))}`);
+            for (const slot of mm.recommended) {
+                console.log(`  ${chalk.cyan(slot.slot.padEnd(8))} → ${slot.model} (${slot.size})`);
+            }
+            console.log();
+        }
+        // Recommendations
+        console.log(`  ${chalk.bold('Quick Start')}`);
+        console.log(`  ${chalk.dim('─'.repeat(50))}`);
+        for (const rec of hw.recommendations) {
+            console.log(`  ${chalk.white(rec)}`);
+        }
+        console.log();
+        // Quantization info
+        console.log(`  ${chalk.dim('Quantization options: Q2 (smallest) → Q4 (default) → Q8 (best) → F16 (original)')}`);
+        console.log(`  ${chalk.dim('Higher quant = better quality, more RAM. Use: kbot models pull hf:user/repo:file-Q6_K.gguf')}`);
+        console.log();
+    });
     const synthCmd = program
         .command('synthesis')
         .description('Closed-loop intelligence compounding — bridge self-discovery and universe discovery');