wolverine-ai 4.0.2 → 4.0.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wolverine-ai",
-  "version": "4.0.2",
+  "version": "4.0.4",
   "description": "Self-healing Node.js server framework powered by AI. Catches crashes, diagnoses errors, generates fixes, verifies, and restarts — automatically.",
   "main": "src/index.js",
   "bin": {

package/src/brain/embedder.js

@@ -43,14 +43,21 @@ async function embed(text) {
 
   const model = getEmbeddingModel();
   const provider = detectProvider(model);
-  // wolverine-embedding-1 routes through billing proxy, others go direct
   const client = provider === "wolverine" ? getClient("wolverine") : getClient("openai");
 
   const startMs = Date.now();
-  const response = await client.embeddings.create({
-    model,
-    input: text,
-  });
+  let response;
+  try {
+    response = await client.embeddings.create({ model, input: text });
+  } catch (err) {
+    // If wolverine proxy is down (startup, crash loop), fall back to OpenAI direct
+    if (provider === "wolverine" && /ECONNREFUSED|ECONNRESET|ETIMEDOUT|fetch failed|Connection error/i.test(err.message || "")) {
+      const directClient = getClient("openai");
+      response = await directClient.embeddings.create({ model: "text-embedding-3-small", input: text });
+    } else {
+      throw err;
+    }
+  }
 
   const embedding = response.data[0].embedding;
   _trackEmbedding(model, response.usage, Date.now() - startMs, true);
@@ -87,10 +94,17 @@ async function embedBatch(texts) {
   const client = provider === "wolverine" ? getClient("wolverine") : getClient("openai");
 
   const startMs = Date.now();
-  const response = await client.embeddings.create({
-    model,
-    input: uncached,
-  });
+  let response;
+  try {
+    response = await client.embeddings.create({ model, input: uncached });
+  } catch (err) {
+    if (provider === "wolverine" && /ECONNREFUSED|ECONNRESET|ETIMEDOUT|fetch failed|Connection error/i.test(err.message || "")) {
+      const directClient = getClient("openai");
+      response = await directClient.embeddings.create({ model: "text-embedding-3-small", input: uncached });
+    } else {
+      throw err;
+    }
+  }
   _trackEmbedding(model, response.usage, Date.now() - startMs, true);
 
   // Sort by index to maintain order

package/src/core/ai-client.js

@@ -232,7 +232,7 @@ async function aiCall({ model, systemPrompt, userPrompt, maxTokens = 2048, tools
         result = await _chatCall(_getWolverineClient(), { model, systemPrompt, userPrompt, maxTokens, tools, toolChoice });
       } catch (proxyErr) {
         // If billing proxy is down (server crashing), fall back to direct GPU
-        const isConnErr = /ECONNREFUSED|ECONNRESET|ETIMEDOUT|fetch failed/i.test(proxyErr.message || "");
+        const isConnErr = /ECONNREFUSED|ECONNRESET|ETIMEDOUT|fetch failed|Connection error/i.test(proxyErr.message || "");
         const directClient = _getWolverineDirectClient();
         if (isConnErr && directClient) {
           console.log(chalk.yellow("  ⚠️  Billing proxy down — using direct GPU (unbilled)"));
@@ -269,7 +269,7 @@ async function aiCallWithHistory({ model, messages, tools, maxTokens = 4096, cat
       try {
         result = await _chatCallWithHistory(_getWolverineClient(), { model, messages, tools, maxTokens });
       } catch (proxyErr) {
-        const isConnErr = /ECONNREFUSED|ECONNRESET|ETIMEDOUT|fetch failed/i.test(proxyErr.message || "");
+        const isConnErr = /ECONNREFUSED|ECONNRESET|ETIMEDOUT|fetch failed|Connection error/i.test(proxyErr.message || "");
         const directClient = _getWolverineDirectClient();
         if (isConnErr && directClient) {
           console.log(chalk.yellow("  ⚠️  Billing proxy down — using direct GPU (unbilled)"));