β Aura Code
A step-by-step guide for absolute beginners. No technical knowledge required. If you can type and click a mouse, you can use Aura. This manual will take you from zero to AI-powered coding in about 15 minutes.
β Setup Progress Checklist
Tick off each step as you complete it. Your progress is automatically stored in your local storage cache.
π What Is Aura Code?
Your personal AI coding agent that reads, plans, executes, verifies, and reports. No software development degree needed.
In Plain English
Imagine you have a super-smart friend who can look at any folder on your computer and work with its files. That's Aura. You type what you want in plain English, and Aura does it.
aura "add dark mode to my website".Aura reads your files, figures out which ones control colors, edits them, and confirms it works. All you did was type one sentence.
One Architect, Sharp Tools
Aura was designed by DuΕ‘an MilosavljeviΔ β the architecture, the principles, and every decision behind how she works. The code was written with modern coding tools, Claude Code among them.
She works with all major AI models: Claude, GPT, Gemini, MiMo, DeepSeek β and even free local models through Ollama. You pick which brain Aura uses.
Her framework is called Praktess β from Ancient Greek: "she who acts and executes." Her motto: "I don't try. I verify."
βοΈ Installing on Windows 11
Set up Node.js, verify the command prompt environment, and install the Aura CLI wrapper.
Step 1: Download & Install Node.js
Open your web browser
Edge, Chrome, or Firefox β any browser works.
Go to nodejs.org
Type nodejs.org in the address bar at the top and press Enter.
Click the big green LTS button
Click the one that says "LTS" (Long Term Support). A setup installer will download.
Install Node.js
Double-click the downloaded .msi installer file. Accept ALL default settings and click Install. Give Windows admin consent when requested (click Yes).
Step 2: Open Windows Terminal
Press the Windows key on your keyboard. Type cmd or Terminal. Press Enter.
You'll see a command prompt windows. Do not panic! You only need to copy and paste a couple of commands.
Step 3: Verify Installation
Type these commands into the shell and run them:
C:\Users\You> node --version v20.11.0 <-- Node.js is installed successfully C:\Users\You> npm --version 10.2.4 <-- npm package manager is online
Step 4: Install Aura Code Wrapper
In the terminal, run the following global install package instruction:
C:\Users\You> npm install -g aura-code
Wait 1-2 minutes for the compiler download. When completed, verify the aura command is active:
C:\Users\You> aura --version v0.10.5
π Setting Up API Keys
An API key is Aura's password to connect to an AI model brain. Choose a provider and set up credentials.
What Is an API Key?
Think of an API key as a secure token that lets Aura chat with an LLM backend (like Claude or GPT-4). It is a long string of characters typically starting with sk-. Get one from your preferred AI console.
Where to Obtain Keys
| AI Brain Provider | Developer Portal URL | Cost Metrics |
|---|---|---|
| Anthropic (Claude) | console.anthropic.com | Pay-as-you-go (~$3-15/M tokens) |
| OpenAI (GPT-4) | platform.openai.com | Pay-as-you-go (~$1-10/M tokens) |
| Google AI Studio (Gemini) | aistudio.google.com | Free tiers available! |
| Ollama (Local LLMs) | ollama.com | 100% Free / Runs Offline |
How to Configure Environment Keys
Option A: Shell Session Key (active only until you close terminal window):
C:\Users\You> set ANTHROPIC_API_KEY=sk-ant-api03-your-key-here C:\Users\You> set OPENAI_API_KEY=sk-proj-your-key-here
Option B: Permanent Environment Variables (retains after computer reboot):
- Open the Start menu, type
environment. - Click "Edit the system environment variables".
- In System Properties, click "Environment Variables...".
- Under user variables, click "New...".
- Enter Variable Name:
ANTHROPIC_API_KEY(orOPENAI_API_KEY). - Paste your API key value into Variable Value.
- Click OK, apply, and restart Windows Terminal.
aura --models. It lists available brains linked by your keys.
π» Command line logs
Common commands you can type. Remember: always invoke aura first, followed by the instruction.
Everyday Reference Table
| Terminal Command | System Action |
|---|---|
aura "your task description" |
Standard agent loop: Reads code, creates plan, executes, tests, reports. |
aura --help |
List available options, command flags, and help prompts. |
aura --version |
Show installed package version. |
aura --models |
List all online connected model brains. |
aura kanban |
Opens a localhost:7474 web browser server checklist dashboard. |
aura "fix the bug in login controller" |
Finds files, debugs runtime variables, and applies fixes. |
Beginner Prompt Examples
Type these in Windows terminal to test functionality:
$ aura "list all code files in this folder" $ aura "find all TODO tags in the codebase" $ aura "explain the file package.json in simple terms" $ aura "create hello.txt writing 'Verifying agent output' inside"
π§ Agent Personality Modes
Choose the appropriate cognitive architecture settings depending on the size and complexity of your programming task.
Normal Mode (Default)
Ideal for general coding operations: writing features, debugging single files, answering questions.
$ aura "fix class selectors in stylesheet"
Standard loop: Read β Plan β Execute β Verify β Report. Fast and lean.
Orchestrated Mode
Ideal for larger scope projects that require coordinate teams of expert agents.
$ aura --orchestrate "refactor our Express database routing"
Aura splits the task. Separate specialized virtual subagents (Researcher, Architect, Coder, Verifier) execute tasks concurrently.
Architect Mode
Ideal for brainstorming and blueprints. Generates specifications and charts, but does not edit any source files.
$ aura --architect "design a payment microservice"
Verify & Analyze Modes
Verify: Automatically runs test suites and debugs errors until checks pass.
Analyze: Scans system execution history log files to report common failure patterns and optimize token costs.
π₯οΈ Model Brain Index
Aura is model-agnostic. She adapts to whichever brain you feed her. Select the one matching your budget.
| Model Provider | Target Models | Best Target Use Case | Expense Bracket |
|---|---|---|---|
| Claude (Anthropic) | Sonnet 3.5, Opus | Deep reasoning, complex logic | Premium tier ($3-15/M) |
| GPT (OpenAI) | gpt-4o, gpt-4o-mini | Versatile, fast, API integrations | Mid tier ($0.15-5/M) |
| Gemini (Google) | Gemini 2.5 Flash / Pro | Extremely large context (1M+ tokens) | Free options / Low cost |
| MiMo / DeepSeek | MiMo-2.5, DeepSeek Coder | Coding benchmarks, cost savings | Competitive cost |
| Ollama (Local) | Llama 3.2, Qwen 2.5 | Private off-grid coding, offline checks | 100% Free |
π‘οΈ Safety & Permission Protocols
You define the autonomy levels. Aura respects security borders on your operating system.
autonomy levels config
| Safety Level | Operational Permissions |
|---|---|
| read-only | Strictly reads code files. Disables file writes, shell execution, and modifications. Highest safety rating. |
| normal | Can read and edit files locally, but prompts for confirmation before executing shell command tasks. |
| auto | Full autonomous system execution. No permission prompts. Fast but use with caution on trusted projects. |
β System FAQ
Frequently asked questions and terminal diagnostics manual.