@qvac/llm-llamacpp 0.58.0 → 0.59.0

package/README.md

@@ -14,6 +14,8 @@ This native C++ addon, built using the `Bare` Runtime, simplifies running Large
   - [4. Create Model Instance](#4-create-model-instance)
   - [5. Load Model](#5-load-model)
   - [6. Run Inference](#6-run-inference)
+    - [Tool calling](#tool-calling)
+    - [Batch inference](#batch-inference)
   - [7. Release Resources](#7-release-resources)
 - [API behavior by state](#api-behavior-by-state)
 - [Assessing fit](#assessing-fit)
@@ -295,6 +297,26 @@ try {
 
 When `opts.stats` is enabled, `response.stats` includes runtime metrics such as `TTFT`, `TPS`, token counters, and `backendDevice` (`"cpu"` or `"gpu"`). `backendDevice` reflects the resolved device used at runtime after backend selection/fallback logic, not only the requested config.
 
+#### Tool calling
+
+Load with `config.tools: 'true'` and declare tools as `{ type: 'function', ... }` entries in the message array. The model answers a tool request with a tool-call block in its own format (`<tool_call>` for Qwen). To continue the conversation, put the call on an assistant turn as `tool_calls` and the result on a `tool` turn; the chat template renders both natively, so neither needs to be written into `content`.
+
+```javascript
+const response = await model.run([
+  {
+    type: 'function',
+    name: 'get_weather',
+    description: 'Get the weather for a city',
+    parameters: { type: 'object', properties: { city: { type: 'string' } }, required: ['city'] }
+  },
+  { role: 'user', content: 'Weather in Paris?' },
+  { role: 'assistant', content: '', tool_calls: [{ id: 'call_1', name: 'get_weather', arguments: { city: 'Paris' } }] },
+  { role: 'tool', content: '18C', tool_call_id: 'call_1', name: 'get_weather' }
+])
+```
+
+`arguments` is an object or the text of a JSON object, and `content` may be empty on the assistant turn. A malformed tool turn is rejected with `InvalidInputFormat`. [`examples/toolCalling.js`](./examples/toolCalling.js) shows a full round trip.
+
 #### Batch inference
 
 Load the model with `parallel >= 2`, then pass an array of prompts. Each chunk arrives tagged with the id of the sequence that produced it; `await()` returns results in input order.

package/index.d.ts

@@ -375,8 +375,23 @@ declare namespace LlmLlamacpp {
          */
         reasoning_content?: string;
         type?: undefined;
+        /**
+         * Tool calls this assistant turn made, rendered by the chat template in
+         * the model's own tool-call format. `arguments` is an object or the text
+         * of a JSON object. `content` may be empty on such a turn.
+         */
+        tool_calls?: ToolCallTurn[];
+        /** On a `tool` turn: the id of the call this result answers. */
+        tool_call_id?: string;
+        /** On a `tool` turn: the name of the tool that produced the result. */
+        name?: string;
         [key: string]: any;
     }
+    interface ToolCallTurn {
+        id?: string;
+        name: string;
+        arguments?: string | Record<string, unknown>;
+    }
     interface UserMediaMessage {
         role: "user";
         type: "media";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qvac/llm-llamacpp",
-  "version": "0.58.0",
+  "version": "0.59.0",
   "description": "llama addon for qvac",
   "addon": true,
   "scripts": {

package/prebuilds/darwin-arm64/qvac__llm-llamacpp.bare.exports

@@ -7,7 +7,7 @@
     ],
     "current_versions": [
       {
-        "version": "0.58"
+        "version": "0.59"
       }
     ],
     "exported_symbols": [

package/prebuilds/darwin-x64/qvac__llm-llamacpp.bare.exports

@@ -7,7 +7,7 @@
     ],
     "current_versions": [
       {
-        "version": "0.58"
+        "version": "0.59"
       }
     ],
     "exported_symbols": [

package/prebuilds/ios-arm64/qvac__llm-llamacpp.bare.exports

@@ -7,7 +7,7 @@
     ],
     "current_versions": [
       {
-        "version": "0.58"
+        "version": "0.59"
       }
     ],
     "exported_symbols": [

package/prebuilds/ios-arm64-simulator/qvac__llm-llamacpp.bare.exports

@@ -7,7 +7,7 @@
     ],
     "current_versions": [
       {
-        "version": "0.58"
+        "version": "0.59"
       }
     ],
     "exported_symbols": [

package/prebuilds/ios-x64-simulator/qvac__llm-llamacpp.bare.exports

@@ -7,7 +7,7 @@
     ],
     "current_versions": [
       {
-        "version": "0.58"
+        "version": "0.59"
       }
     ],
     "exported_symbols": [