Installation

Download and install Aethaar DevOp on Windows, macOS, or Linux. Covers system requirements, first-run setup, and automated background updates.

System Requirements

Ensure your system meets the following specifications before installing Aethaar DevOp:

Installing the Client

Follow the installer instructions for your operating system:

• Windows: Download the installer (.exe, x64 or ARM64) and run it. Setup completes in one click and adds Start menu and desktop shortcuts.
• macOS: Download the .dmg for Apple Silicon, Intel, or the Universal build, open it, and drag Aethaar DevOp into your /Applications folder.
• Linux: Download the AppImage (x64 or ARM64), make it executable with chmod +x Aethaar-DevOp.AppImage, and run it.

Background Updates

Aethaar DevOp checks for updates shortly after launch. New versions download in the background while the status bar shows progress; when the download finishes, click Restart to update whenever it suits you, or the update installs the next time you quit. If you turn off automatic downloads in Settings, the status bar offers Install Now instead.

Security & First-Time Launch

Quickstart

Open your first project, prompt the AI assistant, and learn the core interface in under ten minutes.

Opening a Project

Launch Aethaar DevOp and choose File → Open Folder, then select the root directory of your project. (Ctrl + O / Cmd + O opens a single file.) DevOp indexes your file names for fast search and reads your manifests (package.json, Cargo.toml, requirements.txt, go.mod) to save your tech stack as a project memory.

Making Your First AI Request

Aethaar provides a responsive AI sidebar directly inside the editor:

• Toggle AI Chat (Ctrl + I / Cmd + I): Open the AI Assistant panel. You can ask architectural questions, request code refactors, or ask the AI to implement multi-file features.
• Context Referencing with @: Type @ in the prompt bar to attach a file or a connected integration (including MCP servers) to your message.
• Autonomous Agent Loops: In Build or Auto mode, the AI can read files, run terminal commands, inspect linter results, and edit code across multiple files. Commands follow your approval setting, and pushes, deploys and other outward or destructive actions always ask first. Ask mode answers questions without changing anything.

Configuration & Sync

Set your preferred AI model, choose a theme, and see what syncs with your Aethaar account.

Accessing Settings

Click the gear icon in the top bar. Settings opens as a tab in the editor area.

Theme & Appearance

Choose Dark, Light, or Auto. Auto follows your operating system's color scheme and switches as soon as it changes. Accent color palettes are available too.

Migration Guide

Transitioning from VS Code, Cursor, or JetBrains to Aethaar DevOp is immediate and frictionless.

Aethaar DevOp operates directly on your local file system without proprietary workspace schemas, locks, or hidden config wrappers. Simply open your existing repository folder to start coding with full syntax highlighting, terminal support, and AI agent assistance.

AI Chat & Context

Ask questions about your codebase, plan architectures, and execute edits with deep natural language understanding.

Context Referencing (@-mentions)

Type @ in the prompt bar to attach context to your message:

• @filename: Attaches that file from your workspace.
• @integration: Attaches a connected integration or MCP server, so the AI knows to use it.

You can also send terminal output to the chat from the terminal panel; it appears as a terminal chip in your message.

Code Search & Context

How the AI finds the right code in your project.

DevOp keeps a local index of your project's file names, refreshed automatically as files change, for instant file search. During a task the AI agent finds code itself with search tools (text search across files, file lookup by name), reads only the files it needs, and can search your project memories and skills. Nothing is uploaded to build an index.

Work Trace & Agent Execution

Real-time visibility into the AI's internal reasoning, planning, and autonomous tool execution.

Expandable Work Trace

While the AI works, each step it takes appears in a collapsible trace above its reply, and the status line shows whether it is thinking, analyzing, working or generating. When the task ends, the trace collapses to Worked for …, which you can expand to review every step:

Inline Terminal & Tool Controls

When the AI runs shell commands or edits files, the output appears inside the chat. Depending on your approval setting you approve or reject each command, and you can open any changed file to inspect its diff. You can stop a task at any time, or send a new message while it works: queue it for after the task, or have the AI pause, summarize its progress, and continue with your message.

Smart Mode (Adaptive Model Routing)

Intelligent routing that automatically selects the most capable and cost-effective AI model for every query.

Routing Strategies

In Smart Mode, DevOp looks at what you are asking (a critical change, deep reasoning, coding, a quick utility task, or a request with images) and picks a suitable model that still has quota:

• Auto Strategy: Balances frontier and standard models according to the task.
• High Strategy: Prefers the strongest reasoning models, such as GPT-6 Astra, Claude Sonnet 5.5 and Qwen 3.8 Max.
• Low Strategy: Uses only fast, economical models, such as Kimi K2.7 Code, GLM 5.3 Flash, DeepSeek V4.1 Flash and GPT-5.6 Luna.
• Custom Pool: Pick your own list of models; Smart Mode chooses only from it.

On the Free plan, Smart Mode skips paid-only models.

Design Mode & Image Understanding

AI image generation and image understanding built directly into your workflow.

Image Generation (Design Mode)

Switch the chat to Design mode to generate visual assets, UI mockups, and diagrams:

• Aspect Ratios: Square (1:1), landscape (16:9, 4:3), and portrait (9:16).
• Several Images at Once: Request 1 to 4 images in a single prompt and download each one.
• Project-Aware Designs: When your prompt refers to this project, app or site, the AI first looks at your project's structure and key files so the mockups match your stack and style.

Image Understanding

Drag and drop screenshots, UI designs, error messages, or diagrams into the chat prompt. Vision-capable models (GPT, Claude, Gemini, Qwen, Kimi, DeepSeek V4.1 Flash and others) can then debug visual bugs, explain diagrams, or turn designs into frontend code. In Smart Mode, a message with an image goes only to a vision-capable model.

Project Memories & Tech Stack Context

Persistent context intelligence that remembers your project conventions and completed tasks across sessions.

Automated Context Detection

When you open a project, DevOp reads its manifests (package.json, Cargo.toml, requirements.txt, go.mod) and saves a Project Tech Stack & Context memory, updating it when the project changes. During tasks the AI saves lasting preferences and project decisions as memories and looks them up when a request depends on earlier work.

Managing Memories

Open the Memories Panel in the AI sidebar to search, edit, or delete memories at any time. Memories are either global (all projects) or tied to one project, and are stored on this device.

Cloud Models & Reasoning Controls

Overview of available Cloud models, interactive preview cards, and fine-grained reasoning effort controls.

Interactive Model Selector

Click the model selector in the chat toolbar to open the combobox. Hovering over any model reveals a live flyout preview card displaying:

• 10-Segment Metric Bars: Visual indicators for Intelligence, Speed, Context Window, and Cost.
• Context Window: From 128K tokens up to about 1M tokens on the largest models.
• Thinking and Speed: Choose the thinking level and, where the provider offers it, a faster speed tier.

Configurable Reasoning & Thinking Effort

Fine-tune how deeply models deliberate before answering. Each model offers only the levels its provider supports:

Speed Tiers

Some providers offer a faster tier at a higher price. Where available, the model's preview card shows a Fast toggle: GPT-6 and GPT-5.6 models (plus Ultrafast on GPT-6 Astra), Claude Opus 5.5, and Gemini 4 Argon.

Cloud Model Lineup Overview

Plan Tiers & Model Access

Understanding model access, token quotas, and tier privileges across Free and Paid plans.

Free Tier Access

Free accounts can use every Cloud model that is not paid-only (for example Gemini 3.8 Flash, DeepSeek V4 Pro and V4.1 Flash, Qwen 3.8 Max, Kimi K3, GLM 5.3 Flash, Claude Haiku 5.5 and GPT-5.6 Luna), with a usage allowance that resets every 6 days. Local models are unlimited on every plan. Bring Your Own Key starts with the Starter plan.

Frontier Paid-Only Models

The most expensive frontier models, such as GPT-6 Astra, GPT-6.1 Sol, GPT-5.6 Sol and Terra, Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.1 and Gemini 4 Argon, are available on the Starter, Pro and Ultimate plans.

Bring Your Own Key (BYOK)

Connect personal API keys for direct model access, zero platform markups, and higher rate limits.

Supported BYOK Providers

BYOK is available on the Starter, Pro and Ultimate plans. Add your API keys in the BYOK section of your Aethaar dashboard (Manage API Keys in Portal in the app), then enable the models you want under Cloud & BYOK in the app's model manager:

• Direct Providers: OpenAI, Anthropic, Google Gemini, DeepSeek, and xAI (Grok).
• OpenRouter: One OpenRouter key gives you models from many providers, including Claude, GPT, Gemini, Grok, Qwen, Kimi, GLM, DeepSeek, Mistral and MiniMax.

BYOK requests are billed by your provider. The Usage Log on your dashboard shows the tokens each task used (including cached input) and an estimated cost at the provider's list price; OpenRouter rows show the exact amount OpenRouter charged.

BYOK Thinking and Speed Controls

Each BYOK model offers the thinking levels its provider supports (for example Claude Max, OpenAI X-High and Max, Grok 4.7 X-High, DeepSeek Low / High / Max). Speed options follow each provider too: Fast on OpenAI GPT-6 and GPT-5.6 models, Claude Opus 5.5 and Opus 4.8 (Opus 4.8 Fast needs access from Anthropic), and Gemini Priority on Gemini 4 Argon, 3.8 Flash, 3.6 Flash, 3.5 Flash-Lite and 3.1 Pro. On OpenRouter models, Fast sends your request to the provider currently serving the model fastest, at no extra charge.

Local Models & Marketplace

Run open-weight models such as Gemma, Qwen, Llama and Mistral on your own computer with DevOp's built-in local engine.

In-App Hugging Face Search

Search Hugging Face for GGUF models from the Local Models tab of the model manager (up to 30 results per search), compare sizes and quantizations, and download a model with one click.

Recommended Models

• Small and fast: Gemma 4 E4B, Qwen 3.5 4B, Phi-3 Mini 4K.
• General purpose: Qwen 3.5 9B, Llama 3 8B Instruct, Mistral 7B v0.3, Gemma 2 9B IT, Qwen 2.5 7B.
• Coding and reasoning: DeepSeek Coder V2 Lite, DeepSeek R1 Distill Qwen 7B.

Running Models Locally

Choose the local model source in the chat toolbar and pick a downloaded model. DevOp runs it with its built-in engine on your machine, so once the model is downloaded your prompts never leave your computer and there are no token costs.

Cloud Provider Integrations

Connect your cloud services so AI agents can inspect schemas, verify deployments, and assist with infrastructure tasks.

Connecting Your Accounts

Open Apps in your Aethaar dashboard to connect your accounts. Supabase, Firebase and Google Cloud connect with OAuth (Google Cloud also accepts a service-account key); Vercel, Netlify, Cloudflare and DigitalOcean connect with an access token. Then link an account to a project from DevOp, so each project can use its own accounts.

The Apps page lists more services (such as DigitalOcean, AWS and Stripe) that you can connect to your account; the AI agent can currently call the APIs of the six services above. GitHub is connected separately from the GitHub page of your dashboard and DevOp's GitHub panel, where you can clone your repositories and commit, push, pull and publish.

AI Agent Tool Access

The AI agent can list the accounts linked to your project and read from their APIs during a task, and every call is shown in the chat. Anything that changes data on a service (a non-read request), and every push, deploy or publish, always asks for your approval first.

Agent Skills & MCP

Extend the AI agent with built-in skill guides and the open Model Context Protocol.

Built-in Skills

Skills are instruction guides the agent searches and loads when a task calls for them. DevOp ships skills for common work, including:

• Codebase Audit: Looks for unnecessary dependencies, duplicate utilities and over-engineered code.
• Debt Ledger: Collects deferred-work markers in your code into one list for later cleanup.
• Minimalist Code Review: Reviews changes with a focus on simplicity and standard-library alternatives.

Model Context Protocol (MCP)

Add MCP servers (local stdio commands or remote HTTP servers) from the MCP panel in the AI sidebar, or import servers you already set up in another editor. Connected servers start with DevOp, and the AI agent can call their tools during a chat; you can also attach a server to a message with @.

Internal Browser Preview & DevTools

Test and preview your web applications directly inside Aethaar DevOp without leaving the editor.

Automatic Port Detection

When you start a development server (e.g. npm run dev, vite, next dev) in the built-in terminal, DevOp spots the local URL it prints (e.g. http://localhost:3000 or http://localhost:5173) and opens it in the internal browser.

DevTools

Open DevTools from the internal browser's toolbar to get the full Chromium developer tools (Elements, Console, Network, Sources and more) for the previewed page.

Default Shortcuts

Comprehensive keybinding reference for Windows, macOS, and Linux.