Skip to content
| Marketplace
Sign in
Visual Studio Code>AI>OCLite Copilot ExtensionNew to Visual Studio Code? Get it now.
OCLite Copilot Extension

OCLite Copilot Extension

oclite.site

|
2 installs
| (0) | Free
AI-powered creative asset generation with Azure cloud integration. Generate game assets, icons, and graphics directly in VS Code using advanced AI models.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

OCLite Copilot Extension

OCLite Logo

AI-powered creative asset generation for Visual Studio Code

VS Code Marketplace License Verified Publisher


Overview

OCLite Copilot Extension brings AI image generation directly into your development workflow. Describe what you need in natural language, and the extension's multi-agent pipeline refines your prompt through GPT-4o mini before generating high-fidelity images — all without leaving VS Code.

New: Multi-Agent Pipeline

OCLite now includes an agentic system that can analyze your code and automatically generate relevant visual assets:

  1. Context Analyzer Agent — Reads your source files and identifies entities that need visual assets.
  2. Creative Prompt Agent — Transforms the analysis into detailed, optimized image-generation prompts.
  3. Orchestrator — Coordinates the agents and delivers results to the UI.

Right-click any file or folder in the Explorer and select "OCLite: Analyze & Generate Assets" to try it.

Architecture

Cloud-Native & Serverless

Component Technology Role
LLM Gateway Azure Functions (Node.js) Serverless proxy to GPT-4o mini
Secrets Management Azure Key Vault Secure API key storage
Image Generation SDXL, Flux, Animagine (via OCLite API) High-fidelity asset generation
UI Framework VS Code Webview UI Toolkit Native VS Code experience
Extension Host TypeScript + VS Code API Agent orchestration & commands

Data Flow

User prompt  OR  Right-click file/folder
       │                    │
       │                    ▼
       │         ┌─────────────────────┐
       │         │ Context Analyzer    │  ← Reads code, extracts entities
       │         │ Agent               │
       │         └──────────┬──────────┘
       │                    │ creative brief
       │                    ▼
       │         ┌─────────────────────┐
       ├────────►│ Creative Prompt     │  ← Generates optimized prompts
                 │ Agent (GPT-4o mini) │
                 └──────────┬──────────┘
                            │ refined prompt
                            ▼
                 ┌─────────────────────┐
                 │ Image Generation    │  ← SDXL Lightning, Flux, etc.
                 │ (OCLite API)        │
                 └──────────┬──────────┘
                            │ image
                            ▼
                 ┌─────────────────────┐
                 │ Save / Preview      │  ← Smart naming, workspace-aware
                 └─────────────────────┘

Project Structure

src/
├── agents/
│   ├── AgentOrchestrator.ts      # Coordinates the multi-agent pipeline
│   ├── ContextAnalyzerAgent.ts   # Reads & analyzes code files
│   └── CreativePromptAgent.ts    # Brief → detailed image prompts
├── services/
│   ├── llm.ts                    # Single LLM gateway (GPT-4o mini)
│   └── ai.ts                    # Prompt refinement & smart naming
├── panels/
│   ├── ChatProvider.ts           # Chat webview + agent request handler
│   └── SidebarProvider.ts        # Image generation sidebar
├── utilities/
│   ├── getNonce.ts               # CSP nonce generation
│   ├── getUri.ts                 # Webview URI helper
│   └── secrets.ts                # XOR crypto + request signing (Layer 1 & 3)
└── extension.ts                  # Entry point, command registration

Features

Chat Participant

Type @oclite in GitHub Copilot Chat to generate images conversationally.

@oclite a futuristic cyberpunk city at night with neon lights

Agent-Powered Analysis

Right-click any file or folder and select "OCLite: Analyze & Generate Assets". The agent pipeline will:

  1. Read and understand your code context.
  2. Identify entities that need visual assets.
  3. Generate optimized prompts automatically.
  4. Display results in the Chat panel.

Sidebar Panel

A dedicated sidebar with a full generation UI, real-time progress feedback, and workspace-aware suggestions.

Multi-Model Support

Model Description
sdxl-lightning Fastest generation (default)
flux-schnell High quality, fast
flux-dev High quality, standard
animagine-xl Anime style specialized
realistic-vision Photorealistic style

Workspace Detection

Automatically detects the project type and suggests the best asset style:

Project Suggested Style
Unity Character / Texture
Unreal Engine Environment / Texture
Godot Pixel Art
React / Vue / Angular UI Icon

Prerequisites

  • Visual Studio Code 1.90.0 or later
  • GitHub Copilot extension (for the @oclite chat participant)

No API key required. OCLite's AI services are fully embedded and ready to use out-of-the-box.

Getting Started

1. Install the Extension

  • Install from Visual Studio Code Marketplace
  • Or install manually: code --install-extension oclite-vscode-x.x.x.vsix

2. Basic Setup (Required)

  1. Set API Key: Run OCLite: Set API Key command to configure your OCLite API key
  2. Start Creating: Use any of the generation methods below

3. Optional Cloud Features (Not Required)

Note: Basic image generation works perfectly without cloud setup!

  • Microsoft Authentication: Run OCLite: Sign in with Microsoft for cloud storage access
  • Azure Blob Storage: Run OCLite: Configure Cloud Storage to enable image sharing
  • Usage Analytics: Run OCLite: Configure Telemetry to set up Application Insights (optional)

4. Generation Methods

  • Chat: Type @oclite <your prompt> in Copilot Chat
  • Sidebar: Select the OCLite icon in the Activity Bar and enter a prompt
  • Agent: Right-click a file or folder → OCLite: Analyze & Generate Assets

🎯 What Works Without Cloud Setup

✅ Generate AI images with multiple models
✅ Save images to your local workspace
✅ Preview images in VS Code editor
✅ Smart prompt enhancement with GPT-4o mini
✅ Multi-agent analysis for code-based asset generation

☁️ What Requires Cloud Setup (Optional)

🔗 Public sharing URLs for images
📱 Cloud image gallery across devices
📊 Usage analytics and telemetry

Extension Settings

Setting Description Default
oclite.apiKey Your OCLite API key —
oclite.model AI model for image generation sdxl-lightning

Commands

Command Description
OCLite: Set API Key Store your OCLite API key securely
OCLite: Sign in with Microsoft Authenticate for cloud storage features
OCLite: Configure Cloud Storage Set up Azure Blob Storage for image sharing
OCLite: Configure Telemetry Set up Application Insights analytics (optional)
OCLite: Analyze & Generate Assets Run the multi-agent pipeline on a file or folder
OCLite: Save Image to Workspace Save a generated image to your project
OCLite: Save Image from URL Save an image from a URL
OCLite: Preview Generated Image Open a generated image in the editor
OCLite: Share Image Link Generate shareable public URL for generated images
OCLite: View Gallery Browse your cloud image gallery
OCLite: View Rate Limit Status Check current usage limits
OCLite: Clear API Key Remove your stored API key

Security & Privacy

OCLite uses enterprise-grade security to protect your data:

🔒 Secure Storage

  • API Keys: Stored in VS Code's secure credential store (never in plain text)
  • Azure Connection Strings: Encrypted using VS Code's secrets API
  • User Data: Isolated per Microsoft account with encrypted paths

☁️ Cloud Features Security

  • Microsoft Authentication: Official VS Code authentication APIs
  • Blob Storage: User-isolated containers with public read URLs for sharing
  • Rate Limiting: 10 requests per minute per user to prevent abuse
  • Auto-cleanup: Temporary files automatically removed

📊 Telemetry (Optional)

  • Privacy First: Anonymous usage analytics only (no prompts, images, or personal data)
  • Respects Settings: Automatically disabled if VS Code telemetry is disabled
  • User Control: Can be configured or disabled individually

🛡️ Legacy Security Model

OCLite also includes a 3-layer security system for API credential protection:

Layer Mechanism Effect
Layer 1 — XOR Encryption URLs and keys stored as encrypted byte arrays No readable strings in source or output
Layer 2 — Webpack Minification Variable name mangling and comment stripping Reverse-engineering protection
Layer 3 — Request Signing Time-based X-OCLite-Sig headers Prevents credential reuse outside extension
  • Verified Publisher — Published as a verified publisher on the VS Code Marketplace.

For more details, see SECURITY.md.

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

Release Notes

0.0.4

  • 3-Layer Security Model — API credentials are now protected by XOR encryption, Webpack obfuscation, and per-request signing. No readable secrets exist in the compiled output.
  • Webpack Build System — Migrated from plain tsc to Webpack + TerserPlugin for minified, mangled production bundles.
  • Zero User Configuration — Users install and use immediately, no API key setup required.

0.0.3

  • 3-Layer Security Model — API credentials are now protected by XOR encryption, Webpack minification with variable mangling, and per-request signing. No readable URL or key exists in the packaged extension.
  • Zero-Config Experience — Users no longer need to input any API key. Install and use immediately.
  • Webpack Build System — Migrated from tsc to Webpack for a single optimized bundle with full minification.
  • Request Signing — Every LLM call now includes a time-based X-OCLite-Sig signature header.

0.0.3

  • Multi-Agent Pipeline — New ContextAnalyzerAgent, CreativePromptAgent, and AgentOrchestrator for code-aware asset generation.
  • Refactored LLM Gateway — Single llm.ts module eliminates code duplication across services.
  • Right-click to Analyze — New context menu command for analyzing files and folders.
  • Cleaner Architecture — Separated concerns into agents/, services/, and panels/.

0.0.1

  • Initial release.
  • Chat participant (@oclite) and sidebar panel.
  • LLM-powered prompt refinement via GPT-4o mini.
  • Multi-model image generation (SDXL Lightning, Flux, Animagine, Realistic Vision).
  • Smart auto-naming via LLM with keyword fallback.
  • Workspace detection for Unity, Unreal, Godot, React, Vue, Angular.

License

This project is licensed under the MIT License.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Use of Microsoft trademarks must follow Microsoft's Trademark & Brand Guidelines. Use of third-party trademarks is subject to those third parties' policies.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft