Screeps AI: First Rewrite and Moving to TypeScript

This is the second part of my journey in Screeps. In the first post, I discussed how building the bot in vanilla JavaScript made large refactors error-prone. The goal was to migrate to TypeScript and build a scalable architecture.

Getting the tooling and type definitions dialed in took some initial effort. The screeps-typescript-starter repository provided an excellent baseline for configuring the build pipeline.

Design of the AI

At the foundation of the AI is a hierarchical Manager architecture that coordinates tasks across the empire:

1. Room Managers: Responsible for high-level room economics, controller levels, and construction.
2. Worker Managers: Supervise creep behavior, assignment queues, and task dispatching.
3. Building Managers: Coordinate infrastructure planning, road placement, and extensions.

Creep Specialization

• Harvester: Dedicated to harvesting energy from sources and depositing into spawns and extensions.
• Upgrader: Focuses exclusively on upgrading the Room Controller (RCL).
• Builder: Constructs placed sites and repairs damaged structures.
• Continuous Harvester: Static mining creeps equipped with maximum WORK parts positioned permanently on energy nodes.

Current AI Features

The bot automatically progresses to RCL 2, constructs 5 extensions, and builds paved roads connecting the spawn to energy sources. Higher-level structures are currently placed manually, with the AI automatically assigning builders to construct them.

Defense

Basic defensive routines are in place. Once a tower is erected, the AI automatically targets hostile creeps spawned by NPC invaders.

Future Plans

The next milestones include automating extension placement at RCL 3 and introducing container-based mining for enhanced energy efficiency.