@aws/ml-container-creator 0.5.0 → 0.6.0

package/src/lib/bootstrap-command-handler.js

@@ -199,6 +199,9 @@ export default class BootstrapCommandHandler {
             if (stackOutputs.BatchS3BucketName) {
                 profileData.batchS3Bucket = stackOutputs.BatchS3BucketName;
             }
+            if (stackOutputs.AdapterS3BucketName) {
+                profileData.adapterS3Bucket = stackOutputs.AdapterS3BucketName;
+            }
             if (stackOutputs.BenchmarkS3BucketName) {
                 profileData.benchmarkS3Bucket = stackOutputs.BenchmarkS3BucketName;
             }
@@ -390,6 +393,9 @@ export default class BootstrapCommandHandler {
                 if (outputs.BatchS3BucketName) {
                     console.log(`  ✅ S3 bucket (batch): ${outputs.BatchS3BucketName}`);
                 }
+                if (outputs.AdapterS3BucketName) {
+                    console.log(`  ✅ S3 bucket (adapters): ${outputs.AdapterS3BucketName}`);
+                }
                 if (outputs.BenchmarkS3BucketName) {
                     console.log(`  ✅ S3 bucket (benchmark): ${outputs.BenchmarkS3BucketName}`);
                 }

package/src/lib/cli-handler.js

@@ -204,7 +204,7 @@ VALIDATION OPTIONS:
 
 MCP OPTIONS:
   --smart                     Enable Bedrock-powered smart mode on all MCP servers
-  --discover                  Enable live registry lookups (e.g. Docker Hub) on MCP servers that support it
+  --no-discover               Disable live registry lookups (HuggingFace API, quota checks) — catalog-only mode
 
 REGISTRY SYSTEM:
   The generator includes built-in registries for frameworks, models, and instance types:

package/src/lib/config-manager.js

@@ -1631,7 +1631,7 @@ export default class ConfigManager {
         if (!mcpServerConfigs || !mcpServerConfigs[serverName]) return null;
 
         const smart = this.options.smart === true;
-        const discover = this.options.discover === true;
+        const discover = this.options.discover !== false;
         const serverConfig = mcpServerConfigs[serverName];
 
         // Build a custom McpClient that passes context through

package/src/lib/mcp-client.js

@@ -32,7 +32,7 @@ class McpClient {
         this.timeout = options.timeout || DEFAULT_TIMEOUT;
         this.parameterMatrix = options.parameterMatrix || {};
         this.smart = options.smart || false;
-        this.discover = options.discover || false;
+        this.discover = options.discover !== undefined ? options.discover : true;
         this._transport = null;
         this._client = null;
         this._diagnosticMessage = null;
@@ -98,10 +98,10 @@ class McpClient {
 
         // Build environment: merge process.env with server-specific env
         // When --smart flag is active, inject BEDROCK_SMART=true for this run
-        // When --discover flag is active, inject MCP_DISCOVER=true for this run
+        // Discover mode is now default; inject DISCOVER_MODE=false only when explicitly disabled
         // Always pass process.env so child processes inherit AWS credentials, profiles, etc.
         const smartEnv = this.smart ? { BEDROCK_SMART: 'true' } : {};
-        const discoverEnv = this.discover ? { MCP_DISCOVER: 'true' } : {};
+        const discoverEnv = this.discover === false ? { DISCOVER_MODE: 'false' } : {};
         const serverEnv = env && Object.keys(env).length > 0 ? env : {};
         const spawnEnv = { ...process.env, ...smartEnv, ...discoverEnv, ...serverEnv };
 

package/src/lib/prompt-runner.js

@@ -1098,9 +1098,9 @@ export default class PromptRunner {
         if (!modelName || modelName === 'Custom (enter manually)') return;
 
         const smart = this.options.smart === true;
-        const discover = this.options.discover === true;
+        const discover = this.options.discover !== false;
 
-        const modeLabel = [smart && '[smart]', discover && '[discover]'].filter(Boolean).join(' ');
+        const modeLabel = [smart && '[smart]', !discover && '[no-discover]'].filter(Boolean).join(' ');
         console.log(`   🔍 Querying instance-sizer${modeLabel ? ` ${modeLabel}` : ''}...`);
 
         try {
@@ -1115,8 +1115,8 @@ export default class PromptRunner {
             const { StdioClientTransport } = await import('@modelcontextprotocol/sdk/client/stdio.js');
 
             const serverArgs = [...(serverConfig.args || [])];
-            if (discover && !serverArgs.includes('--discover')) {
-                serverArgs.push('--discover');
+            if (!discover && !serverArgs.includes('--no-discover')) {
+                serverArgs.push('--no-discover');
             }
 
             const transport = new StdioClientTransport({
@@ -1375,7 +1375,7 @@ export default class PromptRunner {
         if (!mcpServers.includes('base-image-picker')) return;
 
         const smart = this.options.smart === true;
-        const discover = this.options.discover === true;
+        const discover = this.options.discover !== false;
         const framework = frameworkAnswers.framework;
         const modelServer = frameworkAnswers.modelServer;
         const architecture = frameworkAnswers.architecture || frameworkAnswers.deploymentConfig?.split('-')[0];

package/src/lib/prompts.js

@@ -399,9 +399,33 @@ const modelFormatPrompts = [
                 ];
             }
             return [
-                'openai/gpt-oss-20b',
-                'meta-llama/Llama-3.2-3B-Instruct',
+                { type: 'separator', separator: '── Meta Llama ──' },
                 'meta-llama/Llama-3.2-1B-Instruct',
+                'meta-llama/Llama-3.2-3B-Instruct',
+                'meta-llama/Llama-3.1-8B-Instruct',
+                'meta-llama/Llama-3.3-70B-Instruct',
+                { type: 'separator', separator: '── Qwen (Alibaba) ──' },
+                'Qwen/Qwen3-0.6B',
+                'Qwen/Qwen3-1.7B',
+                'Qwen/Qwen3-4B',
+                'Qwen/Qwen3-8B',
+                'Qwen/Qwen3-14B',
+                'Qwen/Qwen3-32B',
+                'Qwen/Qwen2.5-7B-Instruct',
+                'Qwen/Qwen2.5-14B-Instruct',
+                'Qwen/Qwen2.5-32B-Instruct',
+                'Qwen/Qwen2.5-72B-Instruct',
+                { type: 'separator', separator: '── DeepSeek ──' },
+                'deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B',
+                'deepseek-ai/DeepSeek-R1-Distill-Qwen-7B',
+                'deepseek-ai/DeepSeek-R1-Distill-Qwen-14B',
+                'deepseek-ai/DeepSeek-R1-Distill-Qwen-32B',
+                'deepseek-ai/DeepSeek-R1-Distill-Llama-8B',
+                'deepseek-ai/DeepSeek-R1-Distill-Llama-70B',
+                { type: 'separator', separator: '── OpenAI ──' },
+                'openai/gpt-oss-20b',
+                'openai/gpt-oss-120b',
+                { type: 'separator', separator: '──────────────' },
                 'Custom (enter manually)'
             ];
         },
@@ -413,7 +437,7 @@ const modelFormatPrompts = [
             if (architecture === 'diffusors') {
                 return 'stabilityai/stable-diffusion-3.5-medium';
             }
-            return 'openai/gpt-oss-20b';
+            return 'meta-llama/Llama-3.1-8B-Instruct';
         },
         when: answers => {
             const architecture = answers.architecture || answers.deploymentConfig?.split('-')[0];
@@ -528,9 +552,11 @@ const modelProfilePrompts = [
  */
 // eslint-disable-next-line no-unused-vars -- reference list for future use
 const EXAMPLE_MODEL_IDS = [
-    'openai/gpt-oss-20b',
+    'meta-llama/Llama-3.1-8B-Instruct',
     'meta-llama/Llama-3.2-3B-Instruct',
-    'meta-llama/Llama-3.2-1B-Instruct'
+    'Qwen/Qwen3-8B',
+    'deepseek-ai/DeepSeek-R1-Distill-Qwen-7B',
+    'openai/gpt-oss-20b'
 ];
 
 const hfTokenPrompts = [

package/templates/do/adapter

@@ -162,8 +162,8 @@ _download_from_hub() {
 
     # ── Resolve S3 bucket ─────────────────────────────────────────────────
     local s3_bucket=""
-    if [ -n "${ASYNC_S3_BUCKET:-}" ]; then
-        s3_bucket="${ASYNC_S3_BUCKET}"
+    if [ -n "${ADAPTER_S3_BUCKET:-}" ]; then
+        s3_bucket="${ADAPTER_S3_BUCKET}"
     else
         local account_id
         account_id=$(aws sts get-caller-identity --query Account --output text 2>/dev/null || echo "")
@@ -172,14 +172,30 @@ _download_from_hub() {
             echo "   Ensure AWS credentials are configured."
             exit 1
         fi
-        s3_bucket="ml-container-creator-${account_id}-${AWS_REGION}"
+        s3_bucket="mlcc-adapters-${account_id}-${AWS_REGION}"
     fi
 
     # ── Create temp directory ─────────────────────────────────────────────
     mkdir -p "${tmp_dir}/adapter_files"
 
     # ── Download adapter files ────────────────────────────────────────────
-    if command -v huggingface-cli &>/dev/null; then
+    if command -v hf &>/dev/null; then
+        echo "   Using hf CLI to download..."
+        local hf_args=("download" "${hf_repo_id}" "--local-dir" "${tmp_dir}/adapter_files")
+        if [ -n "${HF_TOKEN:-}" ]; then
+            hf_args+=("--token" "${HF_TOKEN}")
+        fi
+        if ! hf "${hf_args[@]}" 2>/dev/null; then
+            echo "❌ Failed to download adapter from HuggingFace Hub: ${hf_repo_id}"
+            echo ""
+            echo "   Check that:"
+            echo "   • The repository exists: https://huggingface.co/${hf_repo_id}"
+            echo "   • For gated repos, set HF_TOKEN environment variable"
+            echo "   • You have network connectivity to huggingface.co"
+            rm -rf "${tmp_dir}"
+            exit 1
+        fi
+    elif command -v huggingface-cli &>/dev/null; then
         echo "   Using huggingface-cli to download..."
         local hf_args=("download" "${hf_repo_id}" "--local-dir" "${tmp_dir}/adapter_files")
         if [ -n "${HF_TOKEN:-}" ]; then
@@ -693,7 +709,7 @@ _adapter_list() {
 
     # ── Print table ───────────────────────────────────────────────────────
     printf '%-14s%-12s%s\n' "NAME" "STATUS" "WEIGHTS"
-    echo -e "${output_lines}" | head -n -1
+    echo -e "${output_lines}" | sed '$ { /^$/d; }'
 }
 
 _adapter_remove() {

package/templates/do/config

@@ -13,6 +13,7 @@ export MODEL_SERVER="<%= modelServer %>"
 <% if (typeof enableLora !== 'undefined' && enableLora) { %>
 # LoRA adapter serving
 export ENABLE_LORA=true
+export ADAPTER_S3_BUCKET="mlcc-adapters-$(aws sts get-caller-identity --query Account --output text 2>/dev/null || echo 'UNKNOWN')-${AWS_REGION}"
 <% } %>
 
 # AWS configuration
@@ -67,7 +68,7 @@ ACCOUNT_ID=$(aws sts get-caller-identity --query Account --output text 2>/dev/nu
 <% if (asyncS3OutputPath) { %>
 export ASYNC_S3_OUTPUT_PATH="<%= asyncS3OutputPath %>"
 <% } else { %>
-export ASYNC_S3_OUTPUT_PATH="s3://ml-container-creator-async-${AWS_REGION}-${ACCOUNT_ID}/${PROJECT_NAME}/output/"
+export ASYNC_S3_OUTPUT_PATH="s3://mlcc-async-${ACCOUNT_ID}-${AWS_REGION}/${PROJECT_NAME}/output/"
 <% } %>
 
 <% if (asyncSnsSuccessTopic) { %>
@@ -107,12 +108,12 @@ ACCOUNT_ID=$(aws sts get-caller-identity --query Account --output text 2>/dev/nu
 <% if (batchInputPath) { %>
 export BATCH_INPUT_PATH="<%= batchInputPath %>"
 <% } else { %>
-export BATCH_INPUT_PATH="s3://ml-container-creator-batch-${AWS_REGION}-${ACCOUNT_ID}/${PROJECT_NAME}/input/"
+export BATCH_INPUT_PATH="s3://mlcc-batch-${ACCOUNT_ID}-${AWS_REGION}/${PROJECT_NAME}/input/"
 <% } %>
 <% if (batchOutputPath) { %>
 export BATCH_OUTPUT_PATH="<%= batchOutputPath %>"
 <% } else { %>
-export BATCH_OUTPUT_PATH="s3://ml-container-creator-batch-${AWS_REGION}-${ACCOUNT_ID}/${PROJECT_NAME}/output/"
+export BATCH_OUTPUT_PATH="s3://mlcc-batch-${ACCOUNT_ID}-${AWS_REGION}/${PROJECT_NAME}/output/"
 <% } %>
 export BATCH_INSTANCE_COUNT="<%= batchInstanceCount %>"
 export BATCH_SPLIT_TYPE="<%= batchSplitType %>"
@@ -223,7 +224,7 @@ export BENCHMARK_REQUEST_COUNT=""
 <% if (benchmarkS3OutputPath) { %>
 export BENCHMARK_S3_OUTPUT_PATH="<%= benchmarkS3OutputPath %>"
 <% } else { %>
-export BENCHMARK_S3_OUTPUT_PATH="s3://ml-container-creator-benchmark-${AWS_REGION}-$(aws sts get-caller-identity --query Account --output text)/${PROJECT_NAME}/"
+export BENCHMARK_S3_OUTPUT_PATH="s3://mlcc-benchmark-$(aws sts get-caller-identity --query Account --output text)-${AWS_REGION}/${PROJECT_NAME}/"
 <% } %>
 export BENCHMARK_JOB_NAME=""
 export BENCHMARK_WORKLOAD_CONFIG_NAME=""