WorkTranslator

Transform casual messages into polished, audience-appropriate professional communication — powered by your own writing style.

WorkTranslator is a VS Code extension that uses GitHub Copilot's language models to rewrite your informal drafts into clear, professional messages tailored for specific audiences (peers, leadership, team members). It learns your personal writing style from sample documents you provide, so every transformation sounds authentically you. The intent here is to use AI to protect your job for once.

Features

Privacy & Data Flow

WorkTranslator sends your draft message and excerpts from your writing samples to the GitHub Copilot language model API for transformation. No data is stored externally — all sample files remain on disk and all processing occurs within your VS Code session. If you enable debug logging, up to 50 characters of input/output may appear in the local output channel.

Getting Started

Prerequisites

• VS Code ≥ 1.85.0
• GitHub Copilot extension installed and activated (provides the language model API)

Quick Start

1. Open the Command Palette (Ctrl+Shift+P / Cmd+Shift+P)
2. Run WorkTranslator: Transform Message
3. Paste your draft, choose an audience, adjust formality, and click Transform

Adding Writing Samples

For best results, provide examples of your professional writing so the extension can learn your style:

1. Create a folder in your workspace (default: WritingSamples/)
2. Add plain-text or markdown files — emails, Slack messages, documents
3. Optionally include YAML frontmatter for richer metadata:

---
type: email
audience: leadership
context: status-update
date: 2024-01-15
description: Weekly security posture update
---

Your email content here...

4. Run WorkTranslator: Reload Samples to pick up new files
5. Run WorkTranslator: Generate Style Profile to rebuild your style profile

Commands

Settings

All settings live under the workTranslator.* namespace. Open Settings (Ctrl+,) and search for "WorkTranslator".

General

Samples & Style

LLM / Model

Debugging

Architecture

WorkTranslator follows a clean modular architecture:

src/
├── extension.ts          # Lean orchestrator — init, wire, dispose
├── commands/             # Command handlers (one file per domain)
│   ├── index.ts          # Command registration
│   ├── serviceContainer.ts
│   ├── transformMessage.ts
│   ├── transformSelection.ts
│   ├── sampleCommands.ts
│   └── styleCommands.ts
├── core/                 # Business logic
│   ├── llmClient.ts      # VS Code Language Model API wrapper
│   ├── messageProcessor.ts  # Transformation orchestrator
│   ├── promptBuilder.ts  # Dynamic prompt construction
│   └── styleEngine.ts    # Style profile generation
├── data/                 # Data & configuration
│   ├── configManager.ts  # Settings access & validation
│   └── sampleManager.ts  # Writing sample loading & caching
├── types/
│   └── index.ts          # All TypeScript interfaces
├── ui/                   # Webview panels & status bar
│   ├── statusBar.ts
│   ├── inputPanel.ts
│   └── outputPanel.ts
└── utils/
    ├── logger.ts
    └── textAnalysis.ts

Tips for Best Results

• Provide 4–8 writing samples across different audiences for accurate style learning.
• Use YAML frontmatter in your samples — it improves audience matching and context selection.
• Start with the default formality (7) and adjust after seeing a few transformations.
• Keep drafts under ~500 words — shorter inputs produce tighter, more focused output.
• Use specific audience commands (Transform for Leadership, etc.) instead of auto-detect when you know your audience.
• Re-generate if the first result doesn't feel right — LLM output varies between runs.

Troubleshooting

Enabling Debug Logs

1. Set workTranslator.enableLogging to true in Settings
2. Run WorkTranslator: Show Logs to open the output channel
3. Reproduce the issue — timestamps, prompts, and errors will appear in the log

Development

# Install dependencies
npm install

# Compile
npm run compile

# Watch (auto-rebuild on changes)
npm run watch

# Launch Extension Development Host
# Press F5 in VS Code

# Run tests (compiles first, then launches VS Code Extension Host)
npm test

Testing

The project includes 90 automated tests across four suites:

Tests run inside a real VS Code Extension Host via @vscode/test-electron to ensure correct interaction with the VS Code API.

License

See LICENSE for details.