@aws/ml-container-creator 0.9.1 → 0.10.3

package/servers/lib/bedrock-client.js

@@ -19,29 +19,29 @@
  * @returns {object|null} Parsed JSON object, or null if extraction fails
  */
 export function extractJson(text) {
-    if (!text || typeof text !== 'string') return null
+    if (!text || typeof text !== 'string') return null;
 
     // Try markdown-fenced code block first (```json ... ``` or ``` ... ```)
-    const fencedMatch = text.match(/```(?:json)?\s*\n?([\s\S]*?)```/)
+    const fencedMatch = text.match(/```(?:json)?\s*\n?([\s\S]*?)```/);
     if (fencedMatch) {
         try {
-            return JSON.parse(fencedMatch[1].trim())
+            return JSON.parse(fencedMatch[1].trim());
         } catch {
             // Fall through to raw extraction
         }
     }
 
     // Try extracting raw JSON object
-    const jsonMatch = text.match(/\{[\s\S]*\}/)
+    const jsonMatch = text.match(/\{[\s\S]*\}/);
     if (jsonMatch) {
         try {
-            return JSON.parse(jsonMatch[0])
+            return JSON.parse(jsonMatch[0]);
         } catch {
-            return null
+            return null;
         }
     }
 
-    return null
+    return null;
 }
 
 /**
@@ -61,22 +61,22 @@ export function extractJson(text) {
  * @returns {Promise<{values: object} | null>}
  */
 export async function queryBedrock(serverConfig, parameters, limit, context) {
-    const prefix = `[${serverConfig.serverName}]`
+    const prefix = `[${serverConfig.serverName}]`;
 
     // Dynamic import with 1s timeout
-    let BedrockRuntimeClient, InvokeModelCommand
+    let BedrockRuntimeClient, InvokeModelCommand;
     try {
         const mod = await Promise.race([
             import('@aws-sdk/client-bedrock-runtime'),
             new Promise((_, reject) =>
                 setTimeout(() => reject(new Error('Import timed out')), 1000)
             )
-        ])
-        BedrockRuntimeClient = mod.BedrockRuntimeClient
-        InvokeModelCommand = mod.InvokeModelCommand
+        ]);
+        BedrockRuntimeClient = mod.BedrockRuntimeClient;
+        InvokeModelCommand = mod.InvokeModelCommand;
     } catch {
-        log(prefix, 'Failed to load @aws-sdk/client-bedrock-runtime. Run "npm install" in the servers/lib/ directory')
-        return null
+        log(prefix, 'Failed to load @aws-sdk/client-bedrock-runtime. Run "npm install" in the servers/lib/ directory');
+        return null;
     }
 
     const client = new BedrockRuntimeClient({
@@ -84,20 +84,20 @@ export async function queryBedrock(serverConfig, parameters, limit, context) {
         requestHandler: {
             requestTimeout: 10000
         }
-    })
+    });
 
     // Build prompt from template
     const contextStr = context && Object.keys(context).length > 0
         ? JSON.stringify(context)
-        : 'No specific configuration context provided.'
+        : 'No specific configuration context provided.';
 
     const prompt = serverConfig.systemPromptTemplate
         .replace('{context}', contextStr)
         .replace('{parameters}', parameters.join(', '))
-        .replace('{limit}', String(limit))
+        .replace('{limit}', String(limit));
 
     try {
-        log(prefix, `Querying Bedrock model ${serverConfig.modelId} in ${serverConfig.region}...`)
+        log(prefix, `Querying Bedrock model ${serverConfig.modelId} in ${serverConfig.region}...`);
 
         const body = JSON.stringify({
             anthropic_version: 'bedrock-2023-05-31',
@@ -107,48 +107,48 @@ export async function queryBedrock(serverConfig, parameters, limit, context) {
                 role: 'user',
                 content: prompt
             }]
-        })
+        });
 
         const command = new InvokeModelCommand({
             modelId: serverConfig.modelId,
             contentType: 'application/json',
             accept: 'application/json',
             body
-        })
+        });
 
-        const response = await client.send(command)
-        const responseBody = JSON.parse(new TextDecoder().decode(response.body))
+        const response = await client.send(command);
+        const responseBody = JSON.parse(new TextDecoder().decode(response.body));
 
-        const text = responseBody.content?.[0]?.text
+        const text = responseBody.content?.[0]?.text;
         if (!text) {
-            log(prefix, 'Bedrock response contained no text content')
-            return null
+            log(prefix, 'Bedrock response contained no text content');
+            return null;
         }
 
-        const parsed = extractJson(text)
+        const parsed = extractJson(text);
         if (!parsed) {
-            log(prefix, 'Could not extract JSON from Bedrock response')
-            return null
+            log(prefix, 'Could not extract JSON from Bedrock response');
+            return null;
         }
 
         if (!parsed.values || typeof parsed.values !== 'object') {
-            log(prefix, 'Bedrock response missing "values" object')
-            return null
+            log(prefix, 'Bedrock response missing "values" object');
+            return null;
         }
 
-        log(prefix, `Bedrock returned recommendations: ${JSON.stringify(parsed.values)}`)
-        return parsed
+        log(prefix, `Bedrock returned recommendations: ${JSON.stringify(parsed.values)}`);
+        return parsed;
     } catch (err) {
         if (err.name === 'AccessDeniedException') {
-            log(prefix, `Access denied. Ensure bedrock:InvokeModel permission for arn:aws:bedrock:${serverConfig.region}:*:inference-profile/${serverConfig.modelId}`)
+            log(prefix, `Access denied. Ensure bedrock:InvokeModel permission for arn:aws:bedrock:${serverConfig.region}:*:inference-profile/${serverConfig.modelId}`);
         } else if (err.name === 'ResourceNotFoundException') {
-            log(prefix, `Model "${serverConfig.modelId}" not found. Set BEDROCK_MODEL env var. Example: BEDROCK_MODEL=global.anthropic.claude-sonnet-4-20250514-v1:0`)
+            log(prefix, `Model "${serverConfig.modelId}" not found. Set BEDROCK_MODEL env var. Example: BEDROCK_MODEL=global.anthropic.claude-sonnet-4-20250514-v1:0`);
         } else if (err.name === 'ThrottlingException') {
-            log(prefix, 'Bedrock rate limit hit. Falling back to static recommendations')
+            log(prefix, 'Bedrock rate limit hit. Falling back to static recommendations');
         } else {
-            log(prefix, `Bedrock query failed: ${err.name}: ${err.message}`)
+            log(prefix, `Bedrock query failed: ${err.name}: ${err.message}`);
         }
-        return null
+        return null;
     }
 }
 
@@ -156,5 +156,5 @@ export async function queryBedrock(serverConfig, parameters, limit, context) {
  * Log to stderr so it doesn't interfere with MCP stdio protocol on stdout.
  */
 function log(prefix, message) {
-    process.stderr.write(`${prefix} ${message}\n`)
+    process.stderr.write(`${prefix} ${message}\n`);
 }

package/servers/lib/catalogs/jumpstart-public.json

@@ -4,63 +4,148 @@
         "family": "falcon",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-generation", "llm"],
-        "description": "Falcon 7B via JumpStart"
+        "tags": [
+            "text-generation",
+            "llm"
+        ],
+        "description": "Falcon 7B via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://huggingface-llm-falcon-40b": {
         "modelId": "jumpstart://huggingface-llm-falcon-40b",
         "family": "falcon",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-generation", "llm"],
-        "description": "Falcon 40B via JumpStart"
+        "tags": [
+            "text-generation",
+            "llm"
+        ],
+        "description": "Falcon 40B via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://meta-textgeneration-llama-2-7b": {
         "modelId": "jumpstart://meta-textgeneration-llama-2-7b",
         "family": "llama-2",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-generation", "llm", "llama-2"],
-        "description": "Llama 2 7B via JumpStart"
+        "tags": [
+            "text-generation",
+            "llm",
+            "llama-2"
+        ],
+        "description": "Llama 2 7B via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://meta-textgeneration-llama-2-13b": {
         "modelId": "jumpstart://meta-textgeneration-llama-2-13b",
         "family": "llama-2",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-generation", "llm", "llama-2"],
-        "description": "Llama 2 13B via JumpStart"
+        "tags": [
+            "text-generation",
+            "llm",
+            "llama-2"
+        ],
+        "description": "Llama 2 13B via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://meta-textgeneration-llama-2-70b": {
         "modelId": "jumpstart://meta-textgeneration-llama-2-70b",
         "family": "llama-2",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-generation", "llm", "llama-2"],
-        "description": "Llama 2 70B via JumpStart"
+        "tags": [
+            "text-generation",
+            "llm",
+            "llama-2"
+        ],
+        "description": "Llama 2 70B via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://model-txt2img-stabilityai-stable-diffusion-v2-1-base": {
         "modelId": "jumpstart://model-txt2img-stabilityai-stable-diffusion-v2-1-base",
         "family": "stable-diffusion",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["image-generation", "diffusion", "stable-diffusion"],
-        "description": "Stable Diffusion v2.1 Base via JumpStart"
+        "tags": [
+            "image-generation",
+            "diffusion",
+            "stable-diffusion"
+        ],
+        "description": "Stable Diffusion v2.1 Base via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://huggingface-text2text-flan-t5-xl": {
         "modelId": "jumpstart://huggingface-text2text-flan-t5-xl",
         "family": "flan-t5",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-generation", "text2text", "flan-t5"],
-        "description": "Flan-T5 XL via JumpStart"
+        "tags": [
+            "text-generation",
+            "text2text",
+            "flan-t5"
+        ],
+        "description": "Flan-T5 XL via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     },
     "jumpstart://huggingface-textembedding-gpt-j-6b": {
         "modelId": "jumpstart://huggingface-textembedding-gpt-j-6b",
         "family": "gpt-j",
         "framework": "huggingface",
         "provider": "jumpstart",
-        "tags": ["text-embedding", "gpt-j"],
-        "description": "GPT-J 6B Embedding via JumpStart"
+        "tags": [
+            "text-embedding",
+            "gpt-j"
+        ],
+        "description": "GPT-J 6B Embedding via JumpStart",
+        "validation_level": "community-validated",
+        "framework_compatibility": {
+            "vllm": ">=0.3.0"
+        },
+        "chat_template": "",
+        "gated": false,
+        "architecture": null
     }
 }

package/servers/lib/catalogs/model-servers.json

@@ -66,18 +66,216 @@
             },
             "notes": "vLLM 0.20.2 adds Gemma 4 support, CUDA 12.9, improved multi-GPU. Requires CUDA compat on drivers < 570.",
             "supportedModelTypes": [
+                "afmoe",
+                "apertus",
+                "arcee",
+                "arctic",
+                "aria",
+                "bagel",
+                "baichuan",
+                "bailing_moe",
+                "bailing_moe_linear",
+                "bamba",
+                "bee",
+                "bert",
+                "bert_with_rope",
+                "blip2",
+                "bloom",
+                "chatglm",
+                "cheers",
+                "clip",
+                "colbert",
+                "colmodernvbert",
+                "colpali",
+                "colqwen3",
+                "colqwen3_5",
+                "commandr",
+                "dbrx",
+                "deepseek_eagle",
+                "deepseek_eagle3",
+                "deepseek_mtp",
+                "deepseek_ocr",
+                "deepseek_ocr2",
+                "deepseek_v2",
+                "deepseek_v4",
+                "deepseek_v4_mtp",
+                "deepseek_vl2",
+                "dots1",
+                "dots_ocr",
+                "ernie",
+                "ernie45",
+                "ernie45_moe",
+                "ernie_mtp",
+                "exaone",
+                "exaone4",
+                "exaone4_5_mtp",
+                "exaone_moe",
+                "exaone_moe_mtp",
+                "extract_hidden_states",
+                "fairseq2_llama",
+                "falcon",
+                "falcon_h1",
+                "flex_olmo",
+                "funasr",
+                "fuyu",
                 "gemma",
                 "gemma2",
                 "gemma3",
+                "gemma3_mm",
+                "gemma3n",
+                "gemma4",
+                "gemma4_mm",
+                "glm",
+                "glm4",
+                "glm4_1v",
+                "glm4_moe",
+                "glm4_moe_lite",
+                "glm4_moe_lite_mtp",
+                "glm4_moe_mtp",
+                "glm4v",
+                "glm_ocr",
+                "glm_ocr_mtp",
+                "glmasr",
+                "gpt2",
+                "gpt_bigcode",
+                "gpt_j",
+                "gpt_neox",
+                "gpt_oss",
+                "granite",
+                "granitemoe",
+                "granitemoehybrid",
+                "granitemoeshared",
+                "gritlm",
+                "grok1",
+                "h2ovl",
+                "hunyuan_v1",
+                "hy_v3",
+                "hy_v3_mtp",
+                "hyperclovax",
+                "hyperclovax_vision",
+                "hyperclovax_vision_v2",
+                "internlm2",
+                "internlm2_ve",
+                "internvl",
+                "iquest_loopcoder",
+                "isaac",
+                "jais",
+                "jais2",
+                "jamba",
+                "jina",
+                "jina_vl",
+                "kanana_v",
+                "keye",
+                "kimi_audio",
+                "kimi_k25",
+                "kimi_linear",
+                "kimi_vl",
+                "lfm2",
+                "lfm2_moe",
+                "lfm2_vl",
                 "llama",
+                "llama4",
+                "llama4_eagle",
+                "llama_eagle",
+                "llama_eagle3",
+                "llava",
+                "longcat_flash",
+                "longcat_flash_mtp",
+                "mamba",
+                "mamba2",
+                "medusa",
+                "midashenglm",
+                "mimo",
+                "mimo_mtp",
+                "mimo_v2_flash",
+                "minicpm",
+                "minicpm3",
+                "minicpm_eagle",
+                "minicpmo",
+                "minicpmv",
+                "minimax_m2",
+                "minimax_text_01",
                 "mistral",
+                "mistral_large_3",
                 "mixtral",
+                "mllama4",
+                "mlp_speculator",
+                "modernbert",
+                "molmo",
+                "molmo2",
+                "mpt",
+                "nano_nemotron_vl",
+                "nemotron",
+                "nemotron_h",
+                "nemotron_h_mtp",
+                "nemotron_nas",
+                "nemotron_vl",
+                "nvlm_d",
+                "olmo",
+                "olmo2",
+                "olmo_hybrid",
+                "olmoe",
+                "opencua",
+                "openpangu",
+                "openpangu_mtp",
+                "opt",
+                "orion",
+                "ouro",
+                "ovis",
+                "ovis2_5",
+                "param2moe",
+                "persimmon",
+                "phi",
+                "phi3",
+                "phi3v",
+                "phi4mm",
+                "phi4siglip",
+                "phimoe",
+                "pixtral",
+                "plamo2",
+                "plamo3",
+                "qwen",
                 "qwen2",
+                "qwen2_moe",
+                "qwen2_rm",
+                "qwen2_vl",
                 "qwen3",
+                "qwen3_5",
+                "qwen3_5_mtp",
+                "qwen3_asr_realtime",
+                "qwen3_dflash",
                 "qwen3_moe",
-                "deepseek_v3",
-                "phi3",
-                "command-r"
+                "qwen3_next",
+                "qwen3_next_mtp",
+                "qwen3_vl",
+                "qwen_vl",
+                "rnj1",
+                "roberta",
+                "rvl",
+                "sarvam",
+                "seed_oss",
+                "siglip",
+                "skyworkr1v",
+                "smolvlm",
+                "solar",
+                "stablelm",
+                "starcoder2",
+                "step1",
+                "step3_text",
+                "step3_vl",
+                "step3p5",
+                "step3p5_mtp",
+                "step_vl",
+                "tarsier",
+                "telechat2",
+                "teleflm",
+                "terratorch",
+                "transformers",
+                "ultravox",
+                "voxtral",
+                "voxtral_realtime",
+                "whisper",
+                "zamba2"
             ]
         },
         {