🚀 AI Coding Config Engine

Stop burning API tokens. Squeeze context usage by up to 99.8% using Graphify & RTK! A unified, centralized configuration hub and interactive control center for Claude Code, Codex CLI, and Antigravity CLI (agy). Manage custom agents, reusable skills, rules (ECC standards), MCP servers, and environment settings in one place.

---

⚡ Quick Start

1. Global Setup (Run Once)

Clone the repository and link your configurations globally:

git clone [email protected]:your-github-user/ai-coding-config.git ~/projects/ai-coding-config
cd ~/projects/ai-coding-config
./install.py

Tip: This automatically installs the global ai-config CLI wrapper in ~/.local/bin/ (make sure this is in your PATH). You can re-run ai-config at any time to refresh configurations. Add the --force flag to run completely non-interactively and auto-overwrite any file conflicts (e.g. ./install.py --force).

Native Windows: use docs/WINDOWS.md. Windows and WSL are separate installations; native Windows wrappers use %LOCALAPPDATA%\ai-coding-config\bin, while WSL/Linux wrappers use ~/.local/bin.

🚀 Standalone Antigravity CLI (agy) Setup

If you only want to install the agy quota status checker wrapper and manage Gemini accounts without setting up the entire config engine:

python3 install-agy.py

• Ubuntu / Linux: Add export PATH="$HOME/.local/bin:$PATH" to your ~/.bashrc or ~/.zshrc if not already present, then run source ~/.bashrc.
• Windows: Make sure your Windows PATH environment variable includes %LOCALAPPDATA%\ai-coding-config\bin.
• Verify Globally: Run agy status from any directory.
• Manage Accounts: Run agy account list, agy account add, and agy account use <target>.
• Full CLI Guide: See docs/AGY_CLI.md.

2. Initialize a Project in 1-Second

Navigate to any project directory and run:

• Offline AST Mode (Default):

  ai-config init
  

  Links active AI assistants (claude, agy, codex), sets up project guidelines (CLAUDE.md, etc.), and configures background git hooks for offline AST updates.

• AI Semantic Mode (Deep Extraction):

  ai-config init ai
  

  Performs deep semantic indexing. By default, this runs in zero-API-key mode using your Antigravity CLI (Gemini) quota (gemini-cli backend).

🚀 Optimization Tips (Speed Boost for AI Mode)

To run AI semantic extraction up to 10x faster, you can enable parallel execution, increase token budget, or use a faster model:

# Enable parallel processing & set model to Gemini 3.5 Flash (Low)
export GRAPHIFY_GEMINI_CLI_PARALLEL=1
export GRAPHIFY_GEMINI_CLI_MODEL="Gemini 3.5 Flash (Low)"

# Run extraction with 4 workers and a larger chunk budget
ai-config init ai --max-concurrency 4 --token-budget 120000

---

🛠️ Practical Usage

🖥️ Interactive Web Dashboard

Manage your configuration, toggle active CLI targets, edit rules, or view staged changes in real-time:

./run-web.sh
# Server runs at http://127.0.0.1:8000

🧠 Token-Saving (RTK + Graphify)

Instead of dumping full files and burning thousands of tokens, redirect broad queries to highly optimized codebase subgraphs:

• Update the Graph (usually automated by git hooks):

  graphify update .
  

  Tip: Add the --force flag to force a complete, clean rebuild of the graph database from scratch (e.g., graphify update . --force).
• Query Codebase Architecture/Context:
  • Architecture Overview: rtk graphify query "How does the auth module work?"
  • Code Relationships: rtk graphify path "src/main.py" "src/auth.py"
  • Concept Explanation: rtk graphify explain "Louvain Clustering"
  • Impact Analysis: rtk graphify affected "db_connect"