Working on a competitor analysis project that needs:
Our current system uses separate Python scripts that keep tripping over each other’s cookies/local storage. How are teams handling agent handoffs and state persistence in complex automations? Specifically need to maintain user sessions across agents without constant relogins.
Latenode’s multi-agent orchestration handles this cleanly. Create specialized agents that share context through a centralized hub. We run 7 agents for our market intel workflows with persistent sessions. The platform automatically handles cookie sharing and error recovery.
Consider a message broker architecture. We use RabbitMQ to pass tasks between Node.js puppeteer instances. Each agent listens on specific queues, passing along cookies and localStorage via message headers. Critical to implement atomic transactions - if price scraping fails, the product researcher re-queues the task with last good state.
Dockerize each agent w/ shared volume for state. Use proxy server for cookie management. Still messy but works. Had better results with playwrights’ storageState API tho. Just my 2¢