feat(server): allow Ollama custom via OLLAMA_HOST env var

Update README to specify pulling the `qwen3-embedding:0.6b` model and document the optional `OLLAMA_HOST` environment variable. Adjust server code to import `EMBEDDING_MODEL`, use it for Ollama embeddings, and pass `process.env.OLLAMA_HOST` as the base URL.

Readme.md

@@ -12,7 +12,11 @@ MCP server for exploring the ECMAScript specification and its implementation in
 ## Prerequisites
 
 - **Bun**: Version 1.0.0 or higher ([install](https://bun.sh/docs/installation))
-- **Ollama**: Installed locally with an embedding model (e.g., `qwen3-embedding:0.6b`)
+- **Ollama**: Installed locally with `qwen3-embedding:0.6b` model pulled (`ollama pull qwen3-embedding:0.6b`)
+
+### Environment Variables
+
+- `OLLAMA_HOST` (optional): Ollama server URL. Defaults to `http://localhost:11434`
 
 ## Installation
 

src/mcp-server.ts

@@ -30,7 +30,7 @@ import {
   sectionContentToolMetadata,
   sectionContentToolName,
 } from "./agent-tools/index.js";
-import { STORAGE_DIR } from "./constants.js";
+import { EMBEDDING_MODEL, STORAGE_DIR } from "./constants.js";
 
 // #region MCP Types
 
@@ -75,8 +75,10 @@ export interface SearchSpecMCPOutput extends McpToolOutputBase {
 // #endregion
 
 // Initialize embeddings
+// OLLAMA_HOST env var is optional - @langchain/ollama defaults to http://localhost:11434
 const embeddings = new OllamaEmbeddings({
-  model: "qwen3-embedding:0.6b",
+  model: EMBEDDING_MODEL,
+  baseUrl: process.env.OLLAMA_HOST,
 });
 
 async function main() {