SmythOS Studio: Open-Source Visual AI Agent Builder and deployable runtime stack from SmythOS. Start with an intuitive drag-and-drop workspace, extend with custom code, and deploy your agents anywhere — local, cloud, or edge — with full governance and control.
JavaScript a2aagentic-aiai-agentsdevtoolsmcp-clientn8n-alternativeno-codeopen-sourceorchestrationsmythosvisual-builder
import axios from 'axios';
import { LLMService } from './LLMHelper/LLMService.class';
import config from '../config';
const memCache = new Map<string, number>();
const MODELS_HASH_CACHE_KEY = '__models_hash';
const MODELS_POLLING_INTERVAL = 10000;
const llmService = new LLMService();
export class ModelsPollingService {
private interval: NodeJS.Timeout | null = null;
public async start() {
// Ignore if already running (interval exists)
if (this.interval) {
return;
}
this.interval = setInterval(async () => {
await this.refreshModelsIfChanged();
}, MODELS_POLLING_INTERVAL);
console.log('🟢 ModelsPollingService: started');
}
public stop() {
if (this.interval) {
clearInterval(this.interval);
this.interval = null;
}
console.log('🔴 ModelsPollingService: stopped');
}
private async refreshModelsIfChanged() {
const hasModelsChanged = await this.hasModelsChanged();
if (hasModelsChanged) {
const latestModels = await llmService.getFreshStandardModels();
if (latestModels && Object.keys(latestModels).length > 0) {
await llmService.cacheModels(latestModels, config.cache.STANDARD_MODELS_CACHE_KEY);
console.log('♻️ ModelsPollingService: refreshed');
}
}
}
private async hasModelsChanged() {
try {
const url = `${config.env.API_SERVER}/models`;
const result = await axios.head(url, { headers: { 'x-smyth-debug': true } });
const currentHash = memCache.get(MODELS_HASH_CACHE_KEY);
const newHash = result.headers['x-models-hash'];
if (currentHash !== newHash) {
console.log('🟡 ModelsPollingService: change detected');
memCache.set(MODELS_HASH_CACHE_KEY, newHash);
return true;
}
return false;
} catch (error) {
return false;
}
}
}