diff --git a/src/core/llm/LlmClient.ts b/src/core/llm/LlmClient.ts
new file mode 100644
index 0000000..03616da
--- /dev/null
+++ b/src/core/llm/LlmClient.ts
@@ -0,0 +1,50 @@
+import OpenAI from 'openai';
+
+export interface ChatMessage {
+  role: 'system' | 'user' | 'assistant';
+  content: string;
+}
+
+export interface ChatOptions {
+  model: string;
+  temperature?: number;
+  maxTokens?: number;
+}
+
+export interface ChatResult {
+  content: string;
+  usage?: { promptTokens?: number; completionTokens?: number; totalTokens?: number };
+}
+
+export class LlmClient {
+  private readonly client: OpenAI;
+
+  constructor(opts: { baseUrl: string; apiKey?: string }) {
+    this.client = new OpenAI({ baseURL: opts.baseUrl, apiKey: opts.apiKey || 'not-needed' });
+  }
+
+  async chat(messages: ChatMessage[], options: ChatOptions): Promise<ChatResult> {
+    const response = await this.client.chat.completions.create({
+      model: options.model,
+      temperature: options.temperature,
+      max_tokens: options.maxTokens,
+      messages,
+    });
+
+    const choice = response.choices[0];
+    if (!choice) {
+      throw new Error('LLM returned no choices');
+    }
+
+    return {
+      content: choice.message?.content ?? '',
+      usage: response.usage
+        ? {
+            promptTokens: response.usage.prompt_tokens,
+            completionTokens: response.usage.completion_tokens,
+            totalTokens: response.usage.total_tokens,
+          }
+        : undefined,
+    };
+  }
+}
diff --git a/test/core/llm/LlmClient.test.ts b/test/core/llm/LlmClient.test.ts
new file mode 100644
index 0000000..6c73017
--- /dev/null
+++ b/test/core/llm/LlmClient.test.ts
@@ -0,0 +1,43 @@
+import { describe, expect, it, vi } from 'vitest';
+
+const createMock = vi.fn();
+
+vi.mock('openai', () => ({
+  default: vi.fn().mockImplementation(() => ({
+    chat: { completions: { create: createMock } },
+  })),
+}));
+
+import { LlmClient } from '../../../src/core/llm/LlmClient';
+
+describe('LlmClient', () => {
+  it('sends messages/options to the OpenAI-compatible endpoint and returns content + usage', async () => {
+    createMock.mockResolvedValue({
+      choices: [{ message: { content: 'hello' } }],
+      usage: { prompt_tokens: 10, completion_tokens: 5, total_tokens: 15 },
+    });
+
+    const client = new LlmClient({ baseUrl: 'http://localhost:11434/v1', apiKey: 'key' });
+    const result = await client.chat([{ role: 'user', content: 'hi' }], { model: 'llama3', temperature: 0.7 });
+
+    expect(result).toEqual({
+      content: 'hello',
+      usage: { promptTokens: 10, completionTokens: 5, totalTokens: 15 },
+    });
+    expect(createMock).toHaveBeenCalledWith({
+      model: 'llama3',
+      temperature: 0.7,
+      max_tokens: undefined,
+      messages: [{ role: 'user', content: 'hi' }],
+    });
+  });
+
+  it('throws when the response has no choices', async () => {
+    createMock.mockResolvedValue({ choices: [] });
+
+    const client = new LlmClient({ baseUrl: 'http://localhost:11434/v1' });
+    await expect(client.chat([{ role: 'user', content: 'hi' }], { model: 'llama3' })).rejects.toThrow(
+      /no choices/
+    );
+  });
+});