agent-creator

This approach automates the design and deployment of composable AI agent systems within NestJS projects. It standardizes the orchestration of planners, executors, and evaluators to accelerate building sophisticated agent architectures.

State persistence for AI agent runs in NestJS works by persisting agent runs and plan data in PostgreSQL. It uses TypeORM entities to manage state persistence and handle nested agent calls during execution.

Yes, you can build a research agent in a NestJS application using this method. It provides a modular module structure with standardized prompts, evaluators, and tool registries to enable end-to-end agent orchestration for research tasks.

The best way to structure an orchestrator and executor in NestJS is to use a modular module structure that separates these components. This allows for DAG-based planning, standardized tool registries, and robust dependency handling.

Yes, you need PostgreSQL to handle nested agent calls and tool integration. The architecture relies on PostgreSQL-backed state persistence to store agent runs and plan data during execution.