KERNAX DEVELOPER DOCUMENTATION

Documentation

Kernax is an enterprise-grade AI-native Integrated Development Environment (IDE) built to analyze entire codebases, orchestrate multi-step autonomous coding workflows, and operate with zero source code retention.

⚡ Autonomous Agent

Spawns local terminal, edits files, runs tests, and debugs errors end-to-end.

🧠 Remote Brain

Eliminates personal API key friction with high-throughput cloud inference.

🛡️ Zero Data Retention

Code exists solely in volatile memory. No weights training on your proprietary software.

1. Installation

Kernax is cross-platform and available for Windows, macOS, and Linux.

Windows (x64)

Download the official installer (Kernax-Setup.exe) and launch it. Kernax automatically configures Windows URL protocol handlers and desktop integration.

macOS (Apple Silicon & Intel)

Download the .dmg file, open it, and drag Kernax.app to your /Applications directory. On first launch, open through Finder if prompted by Gatekeeper.

Linux (.AppImage / .deb)

# For Debian/Ubuntu (.deb)
sudo dpkg -i kernax-linux-amd64.deb

# For AppImage
chmod +x Kernax-linux-x86_64.AppImage
./Kernax-linux-x86_64.AppImage

2. Quickstart & First Launch

Launch Kernax IDE on your computer.
Click Giriş Yap in the top-right title bar or open the command palette.
A browser window will open at kernax.com/ide_auth.php. Confirm your account.
The browser instantly sends a cryptographic bridge token back to the IDE via the kernax://auth deep link.
Your user profile, trial balance, and model access are immediately active!

3. Agent Chat (Ctrl+L / Cmd+L)

Press Ctrl+L to open the Kernax AI Chat panel. Unlike standard autocomplete, the Agent mode possesses autonomous execution capabilities:

Full Workspace Awareness: It scans the active git repo, analyzes directory structures, and performs vector semantic queries.
Automated Tool Calling: It can read files, write diff blocks, execute commands in terminal, and run tests.
Review Before Acceptance: All file modifications are presented as visual split diffs. You can Accept or Reject each change with one click.

4. Inline Code Generation (Ctrl+K / Cmd+K)

Select any block of code in the editor and press Ctrl+K. Type an instruction in plain language (e.g., "Refactor this async function to handle timeout and retry with exponential backoff"). Kernax will stream inline diffs directly into your active editor window.

5. Fast Tab Autocomplete (FIM)

As you type in any supported language (TypeScript, Python, Go, Rust, C++, PHP, HTML/CSS), Kernax calculates Fill-in-the-Middle (FIM) completions in milliseconds. Press Tab to accept the suggestion.

6. Context Engine & @Mentions

In the chat prompt, use special symbols to explicitly inject context into the LLM context window:

@file: Reference any specific file in your project.
@code: Reference a symbol, function, or class across workspace files.
@docs: Pull official framework documentation directly into the model context.
@web: Instruct the agent to perform live internet research for fresh API documentation.

7. Project-Specific Rules (`.kernax/rules`)

Create a folder named .kernax in your repository root with a rules file or AGENTS.md. Kernax automatically injects these guidelines into every agent prompt:

# Project Architecture Guidelines
- Always use TypeScript with strict null checks.
- Use Tailwind CSS tokens, never ad-hoc hex values.
- Write Vitest unit tests for all new service functions.
- Maintain zero lint warnings in CI.

8. Security & Zero-Retention Enterprise Compliance

Enterprise Privacy Guarantee: Your proprietary source code is NEVER used to train commercial foundation models. All transmissions between the IDE and Kernax Brain use TLS 1.3 with mutual cryptographic device fingerprinting.

For enterprise on-premise deployments or custom VPC isolated setups, please contact our enterprise team through the Enterprise Portal.