micro-models-agent 0.24.4 → 0.24.5

package/dist/main.js

@@ -4435,12 +4435,12 @@ class OpenAICompatProvider {
     };
     this.rateLimiter = createRateLimiter(config.rateLimits);
   }
-  async* chat(messages, tools) {
+  async* chat(messages, tools, signal) {
     if (!this.rateLimiter.canMakeRequest()) {
       throw new Error(`Rate limit exceeded: ${this.rateLimiter.getConfig().maxRequestsPerMinute} requests per minute`);
     }
     this.rateLimiter.recordRequest();
-    const streamResult = this.doStream(messages, tools);
+    const streamResult = this.doStream(messages, tools, signal);
     let hasToolCall = false;
     let hasText = false;
     let reasoningAcc = "";
@@ -4455,13 +4455,13 @@ class OpenAICompatProvider {
       yield chunk;
     }
     if (!hasToolCall && !hasText) {
-      const fallback = await this.doNonStreaming(messages, tools);
+      const fallback = await this.doNonStreaming(messages, tools, signal);
       for (const chunk of fallback) {
         yield chunk;
       }
     }
   }
-  async* doStream(messages, tools) {
+  async* doStream(messages, tools, externalSignal) {
     const body = {
       model: this.model,
       messages,
@@ -4488,6 +4488,15 @@ class OpenAICompatProvider {
     const controller = new AbortController;
     const totalTimeoutMs = 120000;
     const timeoutId = setTimeout(() => controller.abort(), totalTimeoutMs);
+    if (externalSignal) {
+      if (externalSignal.aborted) {
+        controller.abort();
+      } else {
+        externalSignal.addEventListener("abort", () => controller.abort(), {
+          once: true
+        });
+      }
+    }
     const response = await this.fetchWithRetry(`${this.config.baseUrl}/chat/completions`, {
       method: "POST",
       headers,
@@ -4591,7 +4600,7 @@ class OpenAICompatProvider {
       reader.releaseLock();
     }
   }
-  async doNonStreaming(messages, tools) {
+  async doNonStreaming(messages, tools, externalSignal) {
     const body = {
       model: this.model,
       messages,
@@ -4616,11 +4625,15 @@ class OpenAICompatProvider {
       headers["Authorization"] = `Bearer ${this.config.apiKey}`;
     }
     try {
-      const response = await this.fetchWithRetry(`${this.config.baseUrl}/chat/completions`, {
+      const fetchInit = {
         method: "POST",
         headers,
         body: JSON.stringify(body)
-      });
+      };
+      if (externalSignal) {
+        fetchInit.signal = externalSignal;
+      }
+      const response = await this.fetchWithRetry(`${this.config.baseUrl}/chat/completions`, fetchInit);
       if (!response.ok) {
         const errorText = await response.text();
         throw new Error(t("error.llm_api", {
@@ -9329,7 +9342,7 @@ ${t("tool.truncated", { tokens: Math.ceil(removedChars / 2) })}`;
       const textChunks = [];
       this.emitPhase(iteration, "thinking", onPhase);
       try {
-        for await (const chunk of llmProvider.chat(history, allTools)) {
+        for await (const chunk of llmProvider.chat(history, allTools, this.abortController?.signal)) {
           if (this.shutdownRequested)
             break;
           if (chunk.type === "text" && chunk.content) {
@@ -26256,6 +26269,7 @@ class Repl {
 ${t("repl.interrupt")}
 `));
               this.agent.shutdown();
+              setTimeout(() => process.exit(1), 3000);
             }
             this.lastEscTime = 0;
             return;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "micro-models-agent",
-  "version": "0.24.4",
+  "version": "0.24.5",
   "description": "Micro Models Agent (MMA) — LLM agent harness for small models (Qwen3.5-9B, 32K-64K context)",
   "type": "module",
   "bin": {