Welcome to AA Code's official website at aacode.net

AA Code – AI-Powered Coding Assistant (Kimi, Claude, Gemini, DeepSeek & Codex)

Generate, review, and refactor code using multiple cutting-edge AI models – all from your terminal.

AA Code is a lightweight CLI tool that brings the power of Kimi, Claude, Gemini, DeepSeek, and Codex directly to your command line. Whether you need to quickly prototype a function, explain complex code, or automate repetitive tasks, AA Code lets you choose the best AI for the job without leaving your editor.

📦 Installation

Global Installation (Recommended)

Install globally for direct command-line access:

npm i -g @theone881/aacode

After installation, simply run:

aacode

Build from Source

git clone https://github.com/Theone-best/aacode.git
cd aacode
cargo build --release
# Binary: target/release/aacode

✨ Features

• 🔥 Multi-Model Support – Seamlessly switch between Kimi, Claude, Gemini, DeepSeek, and Codex.
• ⚡ Zero-Config Start – Works out of the box with sensible defaults.
• 📝 Natural Language to Code – Describe what you want, get ready-to-use code snippets.
• 🔍 Code Review & Explanation – Ask aacode to analyze and improve your code.
• 🛠 Flexible Output – Pipe results to files, clipboard, or directly into your editor.
• 🌐 Offline Fallback – (optional) Use local models if available.
• 🧩 Skill System – 10 built-in skills with Anthropic Agent Skills standard format (decision trees, workflows, examples, best practices).
• 🪝 Lifecycle Hooks – Pre/post tool execution hooks via .aacode/hooks.json.
• 📊 Observability – Built-in tracer for model calls, tool calls, and context events.
• 💾 Prompt Cache Tracking – Monitor cache hit rates for cost optimization.
• 🧠 Agent Memory – Hierarchical memory with smart retrieval across turns.
• 🔗 Provider Chain – Automatic failover across multiple API endpoints.

For a full list of commands and options, see the Documentation section.

🧠 Supported Models

| Model | Provider | Best For | |-------|----------|----------| | kimi | Moonshot AI | Fast, cost-effective coding | | claude | Anthropic | Complex reasoning & safety | | gemini | Google | Multimodal & cutting-edge tasks | | deepseek | DeepSeek | High-performance code generation | | codex | OpenAI | Legacy code & compatibility |

You can set a default model via environment variable AACODE_DEFAULT_MODEL.

🚀 Quick Start

First Run

# Launch aacode — first run will guide you through:
# 1. Login (OAuth browser / device code / manual license key)
# 2. API Key setup (choose provider + enter key)
aacode

# Or skip the wizard and configure manually
aacode --init

Interactive Mode

aacode
# Enter prompts interactively with full TUI experience

One-Shot Mode

# Ask a question directly
aacode "explain what this project does"

# Pipe input
echo "refactor this function" | aacode

CLI Options

aacode [OPTIONS] [PROMPT]

Options:
  --cwd <DIR>          Working directory
  --model <MODEL>      Model name (env: AACODE_MODEL)
  --base-url <URL>     API base URL (env: AACODE_BASE_URL)
  --api-key <KEY>      API key (env: AACODE_API_KEY)
  --license <KEY>      License key (env: AACODE_LICENSE)
  --max-turns <N>      Max turns per query (default: 50)
  --max-tokens <N>     Max tokens per response (default: 4096)
  -y, --auto           Auto mode (no confirmation prompts)
  --init               Run configuration wizard
  --agent              Start as remote agent (connect to aacode.net)
  --install-agent      Install as systemd service for auto-start
  --profile <NAME>     Activate a named profile (readonly/coding/review/custom)
  --help               Show help

🧩 Built-in Skills

AA Code ships with 10 production-ready skills following the Anthropic Agent Skills standard:

| Skill | Description | |-------|-------------| | /skill explain | Explain code with decision trees (function/module/error/architecture) | | /skill refactor | Refactor guide with patterns and Red-Green-Refactor workflow | | /skill debug | Systematic 5-step debugging (Reproduce → Isolate → Root Cause → Fix → Verify) | | /skill review | Code review checklist with severity tags and output templates | | /skill test | Test generation with naming conventions and table-driven templates | | /skill plan | Implementation planning with risk assessment | | /skill commit | Conventional commit message generation | | /skill optimize | Performance profiling and optimization with benchmark tools | | /skill document | Documentation generation for 6 doc types | | /skill migrate | Code migration guide with incremental strategy |

# List available skills
/skills

# Run a skill
/skill review

# Create a custom skill
/skill create my-skill

Custom Skills

Place SKILL.md files in:

• Project level: .aacode/skills/<name>/SKILL.md
• Personal level: ~/.aacode/skills/<name>/SKILL.md

---
name: my-skill
description: What this skill does and when to use it
allowed-tools:
  - bash
  - file_read
---
# My Skill

Instructions, decision trees, examples...

📖 Documentation

• Full Command Reference
• Contributing Guidelines

🤝 Contributing

Contributions, issues, and feature requests are welcome!

Feel free to check the issues page or submit a PR.

Development

# Clone
git clone https://github.com/Theone-best/aacode.git
cd aacode

# Build
cargo build

# Run tests
cargo test

# Run in dev mode
cargo run -- "explain this codebase"

# Format + lint
cargo fmt
cargo clippy

📄 License

This software is proprietary and confidential. It is made available for free use for personal and commercial purposes, subject to the following restrictions:

• You may use the software as-is in your projects or workflows.
• You may not modify, adapt, translate, reverse engineer, or create derivative works of this software.
• You may not redistribute, sublicense, or sell the software or any modified version.
• You may not use the software to build a competing product or service.

If you wish to modify the software for internal use, you must obtain explicit written permission from the author. Commercial use of modified versions is strictly prohibited.

All rights not expressly granted are reserved by the author.

© theone881. All rights reserved.