❱ VSCode Extension

The Majestix AI Inference Hub VSCode extension brings multi-model AI chat directly into your editor. It provides a sidebar chat panel with access to all 18 models, SSE streaming, context-aware code commands, and real-time credit balance display -- all authenticated via API key.

---

Overview

Feature	Description
Sidebar chat panel	Full conversational interface in the VSCode sidebar
18 models	Access to every model on the platform (Anthropic, OpenAI, Vertex, OpenRouter)
SSE streaming	Real-time token-by-token response display
8 slash commands	Pre-built code actions (explain, refactor, find bugs, etc.)
Credit balance	Live credit display in the status bar
API key auth	Secure authentication via inf_ prefixed API keys

---

Installation

Install the extension from the Visual Studio Code Marketplace:

1. Open VSCode.

2. Go to Extensions (Ctrl+Shift+X / Cmd+Shift+X).

3. Search for Majestix AI Inference Hub.

4. Click Install.

Alternatively, install from the command line:

code --install-extension inference-platform.inference-vscode

---

Setup

The extension requires an API key to authenticate with the Majestix AI Inference Hub API. API keys are created through the web application.

Step 1: Create an API Key

1. Sign in to the web app at inference-web.web.app or inference-web2.web.app.

2. Navigate to Settings > API Keys.

3. Click Create New Key and provide a friendly name (e.g., "VSCode -- MacBook Pro").

4. Copy the generated key immediately. It starts with inf_ and cannot be viewed again after creation.

Step 2: Configure the Extension

1. Open VSCode Settings (Ctrl+, / Cmd+,).

2. Search for Majestix AI Inference Hub.

3. Paste your API key into the API Key field.

Or set it directly in settings.json:

{
  "inference.apiKey": "inf_your_api_key_here"
}

Step 3: Verify Connection

Open the Majestix AI Inference Hub sidebar panel. If the API key is valid, you will see your current credit balance in the status bar and the model selector will populate with all available models.

---

Features

Sidebar Chat Panel

The extension adds a dedicated chat panel to the VSCode sidebar. It supports full multi-turn conversations with any of the 18 available models.

• Type a message and press Enter to send.

• Responses stream in real-time using Server-Sent Events (SSE).

• Markdown formatting is rendered inline, including code blocks with syntax highlighting.

• Conversation history is maintained for the duration of the session.

Model Selector

A dropdown at the top of the chat panel lets you choose from all 18 models:

Provider	Models
Anthropic	Claude Sonnet 4.6, Claude Haiku 4.5, Claude Opus 4.6
OpenAI	GPT-5 Mini, GPT-5.2, GPT-5.2 Codex, CMO Agent
Vertex AI	Gemini 3 Flash, Gemini 3.1 Pro, Gemini 3 Image, GPT-5 Image, Seedream 4.5
OpenRouter	Llama 4 Maverick, DeepSeek V3.2, DeepSeek R1, Qwen3 Coder, Kimi K2.5, Grok 4.1 Fast

If no model is selected, the platform auto-routes your request to the best model based on the content of your message.

SSE Streaming

All responses are streamed token-by-token via Server-Sent Events. The chat panel displays tokens as they arrive, providing immediate feedback. The streaming protocol follows the platform standard:

data: {"type": "chunk", "content": "Hello"}\n\n
data: {"type": "chunk", "content": " world"}\n\n
data: {"type": "done", "model": "claude-sonnet", "credits_used": 1.24}\n\n

When the done event arrives, the credit balance in the status bar updates automatically.

Credit Balance Display

The extension displays your current available credits in the VSCode status bar. The balance is the sum of remaining monthly plan credits and any top-up balance. It refreshes after each request completes and can be manually refreshed via the Inference: Refresh Balance command.

Slash Commands

The extension provides 8 pre-built commands that operate on selected code. Select code in the editor, then invoke a command via the Command Palette (Ctrl+Shift+P / Cmd+Shift+P) or right-click context menu.

Command	Palette Name	Description
Explain Code	Inference: Explain Code	Explains the selected code in plain language
Refactor	Inference: Refactor Code	Suggests refactoring improvements with rewritten code
Find Bugs	Inference: Find Bugs	Analyzes the selection for potential bugs and issues
Write Tests	Inference: Write Tests	Generates unit tests for the selected code
Optimize	Inference: Optimize Code	Suggests performance optimizations
Document	Inference: Document Code	Generates documentation comments (JSDoc, docstrings, etc.)
Ask About Selection	Inference: Ask About Selection	Opens the chat panel with the selected code as context
General Question	Inference: Ask Question	Opens the chat panel for a free-form question

Context-Aware Code Interaction

When you invoke any slash command, the extension automatically sends the selected code as context to the model. This means the model sees:

• The selected code snippet.

• The file language (for language-aware responses).

• The file path (for project-aware context).

For the Ask About Selection command, the selected code is prepended to your question so the model can reference it directly.

---

Configuration

All settings are under the inference namespace in VSCode settings.

Setting	Type	Default	Description
inference.apiKey	string	""	Your Majestix AI Inference Hub API key (starts with inf_).
inference.defaultModel	string	""	Preferred model key (e.g., claude-sonnet). Leave empty for auto-routing.
inference.apiBaseUrl	string	https://inference-api-611798501438.us-central1.run.app	API base URL. Override for local development.

---

Keyboard Shortcuts

The extension registers the following default keyboard shortcuts:

Shortcut	Action
Ctrl+Shift+I / Cmd+Shift+I	Toggle the Inference sidebar panel
Ctrl+Shift+E / Cmd+Shift+E	Explain selected code

Additional commands can be bound to custom shortcuts via File > Preferences > Keyboard Shortcuts.

---

Architecture

The extension is built with TypeScript and bundled with esbuild for fast loading.

Components

Component	Description
Extension Host	TypeScript entry point that registers commands, the sidebar webview provider, and the status bar item.
Sidebar Webview	HTML/CSS/JS panel rendered in the VSCode sidebar. Handles chat UI, model selection, and message display.
API Client	HTTP client module that communicates with the Cloud Run API. Sends requests to /chat, /code, /models, and /usage/me.
SSE Parser	Parses text/event-stream responses and dispatches chunk, tool_use, done, and error events to the webview.
Status Bar	Displays the current credit balance. Updates after each request and on manual refresh.

Communication Flow

User types message in Sidebar Webview
    |
    v
Webview posts message to Extension Host (VS Code message API)
    |
    v
Extension Host calls API Client
    |
    v
API Client sends POST /chat or /code to Cloud Run
    Headers: X-Api-Key: inf_...
    Body: { message, model, stream: true }
    |
    v
SSE Parser reads streaming response
    |
    v
Each chunk is posted back to Sidebar Webview for display
    |
    v
On "done" event: update status bar credit balance

Build

The extension uses esbuild for bundling:

npm run build        # Production build
npm run watch        # Development mode with file watching
npm run package      # Create .vsix package for distribution

Repository

The extension source code is in a separate repository:

Property	Value
Repository	LendefiMarkets/inference-vscode
Branch	master
Local path	../inference-vscode/ (relative to inference-gcp/)
Language	TypeScript
Bundler	esbuild
Target	VSCode 1.85+

---

Troubleshooting

"Authentication failed" error

• Verify your API key starts with inf_ and is correctly pasted in settings.

• Check that the key has not been revoked on the web app's Settings > API Keys page.

• API keys expire after 90 days. Generate a new key if yours has expired.

No models appearing in the selector

• Confirm the extension has a valid API key configured.

• Check your internet connection. The extension fetches the model list from the API on startup.

• Try running Inference: Refresh Balance from the Command Palette to re-establish the connection.

Credit balance shows 0

• Your plan credits may be exhausted for the current billing cycle.

• Purchase a top-up via the web app at Billing > Top Up.

• The status bar updates after each request. Run Inference: Refresh Balance to force an update.