@zhivex-ai/ollama 0.3.13 → 0.3.14

package/README.md

@@ -5,9 +5,37 @@ Ollama adapter for Zhivex AI SDK.
 ## Install
 
 ```bash
-bun add @zhivex-ai/ollama
+bun add @zhivex-ai/sdk @zhivex-ai/ollama
 ```
 
+## Usage
+
+Start Ollama and pull the model before running the example:
+
+```bash
+ollama pull llama3.2
+```
+
+```ts
+import { generateText } from "@zhivex-ai/sdk";
+import { createOllama } from "@zhivex-ai/ollama";
+
+const ollama = createOllama({
+  baseURL: process.env.OLLAMA_HOST ?? "http://localhost:11434"
+});
+
+const result = await generateText({
+  model: ollama("llama3.2"),
+  prompt: "Explain why local inference is useful."
+});
+
+console.log(result.text);
+```
+
+The adapter supports streaming, callable tools, JSON/structured output, image input for compatible models, and text embeddings through `ollama.embeddingModel(modelId)`. It does not expose provider-hosted tools, remote MCP, audio, files, or shared reasoning controls. Exact capabilities still depend on the locally installed Ollama model.
+
+`createOllama()` reads `OLLAMA_HOST` and otherwise defaults to `http://localhost:11434`. An explicit `baseURL` takes precedence.
+
 Repository and full documentation:
 
 - <https://github.com/Zhivex/zhivex-ai-sdk>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@zhivex-ai/ollama",
-  "version": "0.3.13",
+  "version": "0.3.14",
   "description": "Ollama adapter for Zhivex AI SDK.",
   "license": "MIT",
   "type": "module",
@@ -41,7 +41,7 @@
     "access": "public"
   },
   "dependencies": {
-    "@zhivex-ai/core": "^0.20.0",
+    "@zhivex-ai/core": "^1.0.0",
     "zod": "^4.4.3"
   }
 }