Add Ollama generation module for AI-powered maintenance

Provides text generation via Ollama (llama3) for heartbeat operations
like auto-dedupe, auto-tag, auto-summarize, auto-split, and auto-archive.

src/core/search/ollamaGen.ts

@@ -0,0 +1,34 @@
+const OLLAMA_URL = process.env.OLLAMA_URL || 'http://localhost:11434';
+const GEN_MODEL = process.env.OLLAMA_GEN_MODEL || 'llama3';
+
+let _available: boolean | null = null;
+
+export async function isGenAvailable(): Promise<boolean> {
+  if (_available !== null) return _available;
+  try {
+    const res = await fetch(`${OLLAMA_URL}/api/tags`, { signal: AbortSignal.timeout(2000) });
+    if (!res.ok) { _available = false; return false; }
+    const data = await res.json() as { models?: { name: string }[] };
+    _available = !!data.models?.some(m => m.name.startsWith(GEN_MODEL));
+  } catch {
+    _available = false;
+  }
+  return _available;
+}
+
+export async function generate(prompt: string): Promise<string | null> {
+  if (!(await isGenAvailable())) return null;
+  try {
+    const res = await fetch(`${OLLAMA_URL}/api/generate`, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({ model: GEN_MODEL, prompt, stream: false }),
+      signal: AbortSignal.timeout(60000),
+    });
+    if (!res.ok) return null;
+    const data = await res.json() as { response?: string };
+    return data.response?.trim() ?? null;
+  } catch {
+    return null;
+  }
+}