Chapter 02 | Antigravity Environment Setup and Boundary Delineation

In the last episode, we prepared the One-Pager plan for the T-Block project. At this point, you might be tempted to open the terminal and say: "Go ahead and build the project with Next.js as requested."

Stop. You need one last bit of patience. A repository where an AI is allowed to add, delete, change, and query at will in the root directory is a ticking time bomb.

In true Agentic development, "environment setup" means setting rules for the Agent—delineating the system instruction layer from the business code layer at the physical folder level.

---

1. Turning the Project into a "State"

A regular developer creates a folder. An Agent developer creates an "administration." Enter your project root directory. We need to establish these basic physical architectures:

/t-block-monorepo
  ├── .agents/          # <- Your AI management team's "Office"
  │   ├── workflows/    # Pipeline specifications
  │   └── skills/       # AI's "Superpowers" skill tree
  ├── docs/             # Mandatory persistent memory zone (findings.md, etc.)
  ├── web-client/       # Frontend Next.js code
  └── backend-api/      # Backend Nest.js or Prisma code (if separated)

2. Why is Explicit Separation Necessary?

Disaster Scenario 1: Without docs/ and clear scopes, it's extremely easy for the LLM to get confused about token context while generating code. It might delete configs you had in the root directory or merge frontend and backend configurations into a tangled mess.

Standardized Agreement: Before giving any creation commands, you must issue an uncrossable administrative directive to the Agent:

"From now on, business code must only be placed under web-client/. All documentation and reflections regarding the development process must only be stored in the .agents/ and docs/ directories."

3. Hands-on: Awakening the System Architecture

You don't need to mkdir these yourself. This is the perfect first opportunity to "train your Agent."

Feed this to Antigravity/Agent:

"This is a brand-new multi-repo project. Please create a directory structure called .agents/workflows/ and .agents/skills/ in the current root. Also, create a docs/ directory. Do not write any business logic code. Report to me once execution is complete."

You see, by letting the Agent create these directories themselves, the awareness of "this is a project with a self-governance architecture" is planted in its brain. This is far more effective than manually creating the folders and then describing them to the Agent.

In the next session, we will issue the supreme constitution for this empty "Office": the GEMINI.md.