badgr-cli 1.0.32 → 1.0.34

package/README.md

@@ -61,11 +61,11 @@ badgr serve meta-llama/Llama-3.1-8B-Instruct --gpu L40S --region EU
 
 | Flag | Default | Description |
 |------|---------|-------------|
-| `--gpu <type>` | auto | GPU type — Badgr infers from model size if omitted. Options: RTX_4090, L40S, A6000, A100, H100 |
+| `--gpu <type>` | auto | GPU type override — Badgr Auto selects based on model size if omitted |
 | `--image <img>` | — | Serve a custom container instead of a HuggingFace model |
 | `--task <task>` | — | vLLM task override, e.g. `embed` for embedding models |
 | `--env KEY=VALUE` | — | Environment variable (repeatable) |
-| `--tier 1\|2` | 1 | `1` = managed provider routing (default); `2` = marketplace routing, lower cost |
+| `--tier 1\|2` | 1 | `1` = reliable execution (default); `2` = lower-cost burst capacity |
 | `--count <n>` | 1 | Number of GPUs (1–8) |
 | `--region US\|EU\|AU` | — | Optional region preference. If omitted, Badgr chooses best available capacity. |
 | `--max-price <$/hr>` | — | Hard spend cap per GPU-hour |
@@ -83,10 +83,11 @@ badgr run python train.py --gpu A100 --env HF_TOKEN=$HF_TOKEN --max-runtime 60
 
 | Flag | Default | Description |
 |------|---------|-------------|
-| `--gpu <type>` | auto | GPU type — Badgr picks best available if omitted |
+| `--gpu <type>` | auto | GPU type override — Badgr Auto selects if omitted |
+| `--min-vram <GB>` | — | Minimum VRAM in GB — optional constraint for Auto routing |
 | `--image <img>` | python:3.11-slim | Docker image |
 | `--env KEY=VALUE` | — | Environment variable (repeatable) |
-| `--tier 1\|2` | 1 | `1` = managed provider routing (default); `2` = marketplace routing, lower cost |
+| `--tier 1\|2` | 1 | `1` = reliable execution (default); `2` = lower-cost burst capacity |
 | `--count <n>` | 1 | Number of GPUs |
 | `--region US\|EU\|AU` | — | Optional region preference. If omitted, Badgr chooses best available capacity. |
 | `--max-price <$/hr>` | — | Hard spend cap per GPU-hour |
@@ -98,16 +99,18 @@ badgr run python train.py --gpu A100 --env HF_TOKEN=$HF_TOKEN --max-runtime 60
 
 ## Routing
 
-Badgr automatically searches verified GPU capacity and chooses the best eligible route for your GPU type, workload, price cap, and optional region preference.
+Badgr Auto selects the best eligible route based on GPU type, VRAM, availability, region, workload requirements, and reliability. Advanced users can optionally choose an execution tier or hardware constraint.
 
-**Tier 1** (default) — managed provider routing with guaranteed SLAs.
+Most users should use the default Badgr Auto route. Tiers are an optional advanced control.
 
-**Tier 2** — marketplace routing for lower-cost options when Tier 1 capacity is constrained.
+**Tier 1** (default) — Reliable execution. Best for production workloads, model serving, and jobs where startup reliability matters most. Uses managed routing, readiness checks, fallback, and teardown controls.
+
+**Tier 2** — Lower-cost burst execution. An optional advanced control for cost-sensitive workloads. Availability may vary.
 
 Preview before provisioning:
 
 ```bash
-badgr serve mistral-7b --gpu RTX_4090 --dry-run
+badgr serve meta-llama/Llama-3.1-8B-Instruct --dry-run
 ```
 
 ---
@@ -154,15 +157,23 @@ const client = new OpenAI({
 
 ## GPU options
 
-| Flag value | GPU | VRAM | Estimated Badgr price/hr |
-|-----------|-----|------|-------------|
-| RTX_4090 | NVIDIA RTX 4090 | 24 GB | $0.65–0.89 |
-| L40S | NVIDIA L40S | 48 GB | $1.10–1.40 |
-| A6000 | NVIDIA RTX A6000 | 48 GB | $1.05–1.35 |
-| A100 | NVIDIA A100 | 80 GB | $1.20–1.50 |
-| H100 | NVIDIA H100 | 80 GB | $2.80–3.10 |
+Badgr Auto selects the best eligible GPU for your workload. Add `--gpu <type>` or `--min-vram <GB>` only when you need more control.
+
+Available GPU types may vary by region and current capacity. Run `badgr capacity` or use `--dry-run` to confirm availability before provisioning.
+
+| Flag value | GPU | VRAM | Best for |
+|-----------|-----|------|---------|
+| RTX_3090 | NVIDIA RTX 3090 | 24 GB | Dev, inference |
+| RTX_4090 | NVIDIA RTX 4090 | 24 GB | Inference, training, dev |
+| L40S | NVIDIA L40S | 48 GB | Inference, vLLM, embeddings |
+| A100 | NVIDIA A100 | 40–80 GB | Training, inference |
+| H100 | NVIDIA H100 | 80 GB | Large model training |
+
+Additional GPU types may be routable depending on current capacity — check with `badgr capacity`.
+
+Pricing is confirmed before provisioning. Use `--dry-run` to see pricing before committing.
 
-Prices shown are estimated Badgr rates. Final $/GPU-hour is confirmed before provisioning and may vary by GPU type, availability, region, workload, and runtime.
+Full GPU support details: see [GPU_SUPPORT.md](../../GPU_SUPPORT.md) in the repo root.
 
 ---
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "badgr-cli",
-  "version": "1.0.32",
+  "version": "1.0.34",
   "description": "Badgr, run or serve GPU workloads from one command",
   "type": "module",
   "bin": {
@@ -12,11 +12,11 @@
     "test:watch": "vitest"
   },
   "dependencies": {
-    "@inquirer/prompts": "^5.1.0",
+    "@inquirer/prompts": "^8.5.2",
     "chalk": "^5.3.0"
   },
   "devDependencies": {
-    "vitest": "^1.6.0"
+    "vitest": "^4.1.8"
   },
   "engines": {
     "node": ">=18.0.0"

package/src/api.js

@@ -105,7 +105,7 @@ export function runJob(config, body) {
     apiKey: config.apiKey,
     baseUrl: config.baseUrl,
     body,
-    timeoutMs: 30_000,
+    timeoutMs: 75_000,
   });
 }
 
@@ -115,7 +115,7 @@ export function serveModel(config, body) {
     apiKey: config.apiKey,
     baseUrl: config.baseUrl,
     body,
-    timeoutMs: 30_000,
+    timeoutMs: 75_000,
   });
 }
 

package/src/commands/receipts.js

@@ -33,6 +33,10 @@ function printReceipt(r, chalk) {
   if (r.retries !== undefined) console.log(`    ${chalk.bold('retries:')}  ${r.retries}`);
   if (r.route?.ratePerHour) console.log(`    ${chalk.bold('rate:')}     $${r.route.ratePerHour.toFixed(4)}/hr`);
 
+  if (r.failureCategory ?? r.failure_category ?? r.failure_reason) {
+    const cat = r.failureCategory ?? r.failure_category ?? r.failure_reason;
+    console.log(`    ${chalk.bold('failure:')}  ${chalk.red(cat)}`);
+  }
   if (r.status) console.log(`    ${chalk.bold('status:')}   ${r.status}`);
   console.log(`    ${chalk.dim(ts)}`);
   console.log();

package/src/commands/run.js

@@ -2,6 +2,7 @@ import { requireApiKey } from '../config.js';
 import { callApi, terminateDeployment } from '../api.js';
 import { addReceipt, updateReceipt, generateReceiptId } from '../store.js';
 import { normalizeTier, callWithFallback, HIGH_RATE_THRESHOLD } from '../fallback.js';
+import { formatCliError } from '../errors.js';
 
 /**
  * badgr run python train.py                    # gpu=auto, attached
@@ -28,6 +29,7 @@ export function parseRunArgs(args) {
     if (args[i] === '--no-expanded-search') { flags.noFallback = true; i++; continue; }
     if (args[i] === '--max-runtime')     { flags.maxRuntime = parseFloat(args[++i]); i++; continue; }
     if (args[i] === '--max-cost')        { flags.maxCost = parseFloat(args[++i]); i++; continue; }
+    if (args[i] === '--min-vram')        { flags.minVram = parseFloat(args[++i]); i++; continue; }
     if (args[i] === '--env') {
       const kv = args[++i]; i++;
       if (!flags.env) flags.env = [];
@@ -324,9 +326,33 @@ async function attachToJob(config, depId, { chalk, maxRuntimeMs = null, maxCost
   return { status: 'interrupted', reason: 'signal', exitCode: null, runtimeMs: Date.now() - startMs, failureType: null };
 }
 
+// Known badgr run flags — used to detect broken shell line continuation.
+const _KNOWN_RUN_FLAGS = new Set([
+  '--gpu', '--image', '--count', '--region', '--tier', '--max-price', '--name',
+  '--detach', '--fallback', '--no-fallback', '--strict-capacity',
+  '--no-expanded-search', '--max-runtime', '--max-cost', '--min-vram', '--env',
+]);
+
 export async function runCommand(config, args, chalk) {
   const { flags, positional } = parseRunArgs(args);
 
+  // Detect flags that ended up in the command because of broken shell line continuation
+  // (e.g. `\ ` with trailing space instead of `\<newline>`).
+  const misplaced = positional.filter(a => _KNOWN_RUN_FLAGS.has(a));
+  if (misplaced.length > 0) {
+    console.error(chalk.red(`\n  ✗ These look like badgr flags but were treated as command arguments:`));
+    console.error(chalk.red(`    ${misplaced.join(', ')}`));
+    console.error(chalk.dim(''));
+    console.error(chalk.dim('  This usually means a line continuation has a trailing space.'));
+    console.error(chalk.dim('  Use a single line, or end each continued line with \\ and no space after:'));
+    console.error(chalk.dim(''));
+    console.error(chalk.dim('    badgr run python script.py \\'));
+    console.error(chalk.dim('      --gpu RTX_4090 --max-cost 10 --max-runtime 60'));
+    console.error(chalk.dim(''));
+    process.exitCode = 1;
+    return;
+  }
+
   if (positional.length === 0 && !flags.image) {
     console.error(chalk.red('Usage: badgr run <command...>'));
     console.error(chalk.red('       badgr run --image my/image:latest'));
@@ -356,7 +382,6 @@ export async function runCommand(config, args, chalk) {
     process.exitCode = 1;
     return;
   }
-
   const command      = positional.length > 0 ? positional : undefined;
   const cmdStr = command ? command.join(' ') : '';
   const isSmoke = cmdStr.length < 80 && /print\s*\(|['"]hello/i.test(cmdStr);
@@ -368,39 +393,26 @@ export async function runCommand(config, args, chalk) {
   const envObj       = parseEnvFlag(flags.env);
   const effectiveTier = normalizeTier(flags.tier);
 
-  let gpu = flags.gpu ? flags.gpu.toUpperCase().replace('-', '_') : 'auto';
-
-  if (gpu === 'auto') {
-    console.log(chalk.bold('\n⚡ Running GPU job\n'));
-    if (command) console.log(`  ${chalk.bold('Command:')}  ${command.join(' ')}`);
-    if (image)   console.log(`  ${chalk.bold('Image:')}    ${image}`);
-    if (command) {
-      const profKey = inferProfileFromCommand(cmdStr);
-      const prof = _PROFILES[profKey];
-      console.log();
-      console.log(`  ${chalk.bold('Estimated workload:')}      ${prof.label}`);
-      console.log(`  ${chalk.bold('Estimated minimum VRAM:')}  ${prof.vram}`);
-      if (maxCost)          console.log(`  ${chalk.bold('Max cost:')}                $${maxCost.toFixed(2)}`);
-      if (flags.maxRuntime) console.log(`  ${chalk.bold('Max runtime:')}             ${flags.maxRuntime}min`);
-    }
-    console.log();
-  } else {
-    console.log(chalk.bold('\n⚡ Running GPU job\n'));
-    if (command) console.log(`  ${chalk.bold('Command:')}  ${command.join(' ')}`);
-    if (image)   console.log(`  ${chalk.bold('Image:')}    ${image}`);
-    console.log(`  ${chalk.bold('GPU:')}      ${gpu}`);
-    if (flags.maxRuntime) console.log(`  ${chalk.bold('Max runtime:')} ${flags.maxRuntime}min`);
-    if (maxCost)          console.log(`  ${chalk.bold('Max cost:')}    $${maxCost.toFixed(2)}`);
-    if (flags.maxPrice)   console.log(`  ${chalk.bold('Max price:')}   $${flags.maxPrice.toFixed(2)}/hr`);
-    if (detach)           console.log(`  ${chalk.dim('(detached — returns immediately)')}`);
-    if (flags.env?.length) console.log(`  ${chalk.bold('Env:')}      ${flags.env.join(', ')}`);
-    console.log();
+  const gpu = flags.gpu ? flags.gpu.toUpperCase().replace('-', '_') : undefined;
+
+  console.log(chalk.bold('\n⚡ Running GPU job\n'));
+  if (command) console.log(`  ${chalk.bold('Command:')}  ${command.join(' ')}`);
+  if (image)   console.log(`  ${chalk.bold('Image:')}    ${image}`);
+  if (gpu)     console.log(`  ${chalk.bold('GPU:')}      ${gpu}`);
+  else         console.log(`  ${chalk.bold('GPU:')}      ${chalk.dim('auto')}`);
+  if (flags.minVram)    console.log(`  ${chalk.bold('Min VRAM:')}   ${flags.minVram} GB`);
+  if (flags.maxRuntime) console.log(`  ${chalk.bold('Max runtime:')} ${flags.maxRuntime}min`);
+  if (maxCost)          console.log(`  ${chalk.bold('Max cost:')}    $${maxCost.toFixed(2)}`);
+  if (flags.maxPrice)   console.log(`  ${chalk.bold('Max price:')}   $${flags.maxPrice.toFixed(2)}/hr`);
+  if (detach)           console.log(`  ${chalk.dim('(detached — returns immediately)')}`);
+  if (flags.env?.length) console.log(`  ${chalk.bold('Env:')}      ${flags.env.join(', ')}`);
+  console.log();
 
-    if (!detach && !flags.maxRuntime && !maxCost) {
-      console.log(chalk.dim('  Tip: add --max-runtime 60 or --max-cost 5.00 to cap spend automatically'));
-    }
+  if (!detach && !flags.maxRuntime && !maxCost) {
+    console.log(chalk.dim('  Tip: add --max-runtime 60 or --max-cost 5.00 to cap spend automatically'));
   }
 
+
   console.log(chalk.dim('  Finding suitable capacity...'));
   if (process.env.BADGR_DEBUG === '1' || process.env.BADGR_DEBUG === 'true') {
     console.log(chalk.dim(`  API: ${config.baseUrl}`));
@@ -411,7 +423,8 @@ export async function runCommand(config, args, chalk) {
     return {
       command,
       image,
-      gpu: (gpuOverride || gpu),
+      gpu: gpuOverride || gpu || 'auto',
+      ...(flags.minVram ? { min_vram: flags.minVram } : {}),
       gpu_count: flags.count || 1,
       ...(effectiveRegion ? { region: effectiveRegion } : {}),
       max_price_per_hour: flags.maxPrice,
@@ -526,8 +539,8 @@ export async function runCommand(config, args, chalk) {
     shuttingDown = true;
     try {
       await teardown(reason);
-    } catch (err) {
-      console.error(chalk.red(`  Warning: teardown may have failed: ${err.message}`));
+    } catch {
+      console.error(formatCliError('TEARDOWN_FAILED', { deploymentId: dep.deployment_id, receiptId: rcptId }, chalk));
     }
     process.exit(reason === 'interrupted' ? 0 : 1);
   }
@@ -542,10 +555,7 @@ export async function runCommand(config, args, chalk) {
 
   if (dep.status === 'failed') {
     process.removeListener('SIGINT', handleShutdown);
-    console.error(chalk.red('\n  ✗ Container failed to start (infrastructure error).\n'));
-    console.error(chalk.dim('  The backend retried automatically. All attempts failed.'));
-    console.error(chalk.dim(`  Contact support with receipt ID: ${rcptId}`));
-    console.log();
+    console.error(formatCliError('JOB_INFRASTRUCTURE_FAILURE', { receiptId: rcptId }, chalk));
     process.exitCode = 1;
     return;
   }
@@ -597,13 +607,10 @@ export async function runCommand(config, args, chalk) {
 
   if (finalStatus === 'failed' || (exitCode !== null && exitCode !== 0)) {
     if (failureType === 'infrastructure') {
-      console.error(chalk.red(`\n  ✗ Machine failure — this is not your code.\n`));
-      console.error(chalk.dim('  Contact support with your receipt ID for a refund.'));
+      console.error(formatCliError('JOB_INFRASTRUCTURE_FAILURE', { receiptId: rcptId }, chalk));
     } else {
-      console.error(chalk.red(`\n  ✗ Job failed (exit ${exitCode ?? 'unknown'})\n`));
-      console.error(chalk.dim(`  Check logs: badgr logs ${dep.deployment_id}`));
+      console.error(formatCliError('JOB_FAILED', { exitCode, deploymentId: dep.deployment_id }, chalk));
     }
-    console.log();
     process.exitCode = exitCode ?? 1;
     return;
   }

package/src/commands/serve.js

@@ -2,6 +2,7 @@ import { requireApiKey } from '../config.js';
 import { callApi } from '../api.js';
 import { addDeployment, addReceipt, updateReceipt, generateReceiptId } from '../store.js';
 import { normalizeTier, callWithFallback, HIGH_RATE_THRESHOLD } from '../fallback.js';
+import { formatCliError } from '../errors.js';
 
 /**
  * badgr serve meta-llama/Llama-3.1-8B-Instruct
@@ -27,6 +28,7 @@ export function parseServeArgs(args) {
     if (args[i] === '--no-wait')               { flags.noWait = true; i++; continue; }
     if (args[i] === '--max-cost')              { flags.maxCost = parseFloat(args[++i]); i++; continue; }
     if (args[i] === '--health-path')           { flags.healthPath = args[++i]; i++; continue; }
+    if (args[i] === '--check-nodes')           { flags.checkNodes = args[++i]; i++; continue; }
     // All three aliases map to noMarketplaceFallback
     if (args[i] === '--no-fallback' ||
         args[i] === '--strict-capacity' ||
@@ -128,10 +130,61 @@ async function waitForEndpoint(endpointUrl, deploymentId, config, timeoutMs = 5
   return { ready: false, timedOut: true, depFailed: false };
 }
 
+async function validateComfyNodes(baseUrl, nodeList, chalk) {
+  const nodes = nodeList.split(',').map(n => n.trim()).filter(Boolean);
+  if (!nodes.length) return;
+  const objectInfoUrl = baseUrl.replace(/\/v1\/?$/, '') + '/object_info';
+  console.log(chalk.dim(`\n  Checking custom nodes via ${objectInfoUrl}…`));
+  try {
+    const res = await fetch(objectInfoUrl, { signal: AbortSignal.timeout(15_000) });
+    if (!res.ok) {
+      console.log(chalk.yellow(`  ⚠ Could not verify nodes (HTTP ${res.status})`));
+      return;
+    }
+    const available = await res.json();
+    const found = nodes.filter(n => Object.prototype.hasOwnProperty.call(available, n));
+    const missing = nodes.filter(n => !Object.prototype.hasOwnProperty.call(available, n));
+    if (found.length) console.log(chalk.green(`  ✓ Nodes confirmed: ${found.join(', ')}`));
+    if (missing.length) {
+      console.log(chalk.yellow(`  ⚠ Nodes not found: ${missing.join(', ')}`));
+      console.log(chalk.dim('    These may not be installed in this image.'));
+    }
+  } catch (err) {
+    console.log(chalk.yellow(`  ⚠ Node check failed: ${err.message}`));
+  }
+}
+
+// Known badgr serve flags — used to detect broken shell line continuation.
+const _KNOWN_SERVE_FLAGS = new Set([
+  '--gpu', '--image', '--task', '--count', '--region', '--tier', '--max-price',
+  '--name', '--no-wait', '--max-cost', '--health-path', '--check-nodes',
+  '--no-fallback', '--strict-capacity', '--no-expanded-search', '--env',
+]);
+
 export async function serveCommand(config, args, chalk) {
   const { model, flags } = parseServeArgs(args);
   const customImage = flags.image || null;
 
+  // Detect flags that ended up as positional args due to broken shell line continuation
+  // (e.g. `\ ` with a trailing space instead of `\<newline>`).
+  // After parseServeArgs, only the model name should be in positional. Any extra token
+  // that matches a known badgr flag name was not parsed — likely a syntax mistake.
+  const servePositional = args.filter(a => !a.startsWith('-'));
+  const misplaced = servePositional.slice(1).filter(a => _KNOWN_SERVE_FLAGS.has(a));
+  if (misplaced.length > 0) {
+    console.error(chalk.red(`\n  ✗ These look like badgr flags but were treated as arguments:`));
+    console.error(chalk.red(`    ${misplaced.join(', ')}`));
+    console.error(chalk.dim(''));
+    console.error(chalk.dim('  This usually means a line continuation has a trailing space.'));
+    console.error(chalk.dim('  Use a single line, or end each continued line with \\ and no space after:'));
+    console.error(chalk.dim(''));
+    console.error(chalk.dim('    badgr serve meta-llama/Llama-3.1-8B-Instruct \\'));
+    console.error(chalk.dim('      --gpu L40S --max-cost 10'));
+    console.error(chalk.dim(''));
+    process.exitCode = 1;
+    return;
+  }
+
   if (!model && !customImage) {
     console.error(chalk.red('Usage: badgr serve <model>'));
     console.error(chalk.red('       badgr serve meta-llama/Llama-3.1-8B-Instruct'));
@@ -302,9 +355,10 @@ export async function serveCommand(config, args, chalk) {
         timeoutMs: 10_000,
       });
       if (['failed', 'terminated', 'error'].includes(latest.status)) {
-        console.error(chalk.red(
-          `\n  ✗ Deployment failed before endpoint became ready: ${latest.error || latest.status}\n`
-        ));
+        console.error(formatCliError('HEALTH_CHECK_DEPLOY_FAILED', {
+          deploymentId: dep.deployment_id,
+          failReason:   latest.error || latest.status,
+        }, chalk));
         updateReceipt(rcptId, { status: 'failed', failReason: latest.error || latest.status });
         process.exitCode = 1;
         return;
@@ -317,10 +371,10 @@ export async function serveCommand(config, args, chalk) {
     process.stdout.write('\n');
 
     if (healthResult.depFailed) {
-      console.error(chalk.red(
-        `\n  ✗ Deployment failed during startup: ${healthResult.failReason || 'unknown error'}\n` +
-        `  Check logs: badgr logs ${dep.deployment_id}\n`
-      ));
+      console.error(formatCliError('HEALTH_CHECK_DEPLOY_FAILED', {
+        deploymentId: dep.deployment_id,
+        failReason:   healthResult.failReason,
+      }, chalk));
       updateReceipt(rcptId, { status: 'failed', failReason: healthResult.failReason });
       process.exitCode = 1;
       return;
@@ -330,6 +384,11 @@ export async function serveCommand(config, args, chalk) {
     if (!endpointReady) updateReceipt(rcptId, { status: 'health_check_timeout' });
   }
 
+  // ── Custom-node validation (ComfyUI) ─────────────────────────────────────
+  if (flags.checkNodes && endpointReady) {
+    await validateComfyNodes(endpointUrl, flags.checkNodes, chalk);
+  }
+
   // ── Update receipt with final state ───────────────────────────────────────
   updateReceipt(rcptId, {
     status:      endpointReady ? 'ready' : 'starting',