Installation Guide

Getting Started with A.I.M. (Antigravity Native)

Welcome to Actual Intelligent Memory, the sovereign context and working memory layer explicitly designed for the Antigravity IDE environment on Windows.

🚀 The Sovereign Deployment Manual

1. The Environment

The aim-antigravity exoskeleton operates as a File-Based State Engineering bridge for the IDE agent, completely replacing the old A.I.M. Linux CLI wrapper and daemon polling.

Prerequisites:

• Windows 11 (WSL not strictly required, as the ecosystem is natively porting to Windows arrays).
• Python 3.10+ (For autonomic hooks, memory processors, and database generation).
• Git (For the rigid GitOps bridge).

2. Sovereign Infrastructure (Ollama)

A.I.M.'s deep memory (engram.db) uses the Nomic text embedding model for local, sovereign retrieval, ensuring you never send your codebase memory to a corporate vector database API.

# Install Ollama for Windows (Download from ollama.com first)
ollama pull nomic-embed-text

3. A.I.M. Installation

Clone the repository to securely establish the agent's brain framework.

git clone https://github.com/BrianV1981/aim-antigravity.git
cd aim-antigravity
python -m venv venv
.\venv\Scripts\activate
pip install -r requirements.txt

4. Agent Initialization (The /init Workflow)

⚠️ The legacy terminal CLI commands (aim init and aim tui) have been permanently orphaned. Do not run them.

Instead, initialize the agent natively inside Antigravity by issuing the Slash Workflow to the agent:

/init

This instructs the agent to:

1. Verify the Engram DB connection and GitHub Auth.
2. Establish your aim_operator_profile (Tier 3 Immediate Memory).
3. Establish your aim_project_architecture Knowledge Item.

5. Continuous Re-calibration (The /sync Workflow)

As your repository evolves or new architectural constraints are discovered over the lifecycle of the project, do NOT run /init again. Simply trigger the calibration command:

/sync

---

🏗️ The Boot Verification Gate

If the agent has successfully booted inside Antigravity, it will not answer your first question directly. To ensure Epistemic Certainty, the system forces the agent through a Pre-Flight Lockout (HANDOFF.md).

The agent's first output must always be a structured Boot Acknowledgment, proving it has read your Commander's Intent, the active Edge, and the Issue Tracker before generating tokens.

---

"I believe I've made my point." — A.I.M.