⚡ LangShift
AI Code Transpiler for Visual Studio Code
Convert code between 25+ programming languages instantly.
Powered by 6 AI providers, multi-subagent parallel execution, proactive Code Guard analysis, PII scrubbing, smart L1/L2 caching, and compiler auto-fixing. Enterprise-ready and zero-knowledge local AI support.
Install from Marketplace · Report Bug · Request Feature
main.py ──rename──▶ main.java
┌──────────────┐ ┌──────────────────────┐
│ def greet(): │ AI │ public String greet() │
│ return "Hi"│ ────▶ │ { return "Hi"; } │
└──────────────┘ └──────────────────────┘
🌟 Why LangShift?
Most code translators require manual copy-pasting, break imports, or lock you into a single vendor. LangShift is built for production engineering teams.
| Feature |
Generic Translators |
LangShift Engine |
| Workflow |
Copy-paste to web app |
Rename file (e.g. app.py → app.java) or 1-click folder conversion |
| Providers |
Single vendor lock-in |
6 Providers (Anthropic, OpenAI, Gemini, OpenRouter, Ollama, LM Studio) |
| Offline / Air-Gapped |
❌ Requires cloud |
✅ 100% Offline execution via local Ollama or LM Studio |
| Project Speed |
Serial file conversion |
Multi-subagent parallel task execution (3x–5x speedup) |
| Code Accuracy |
Single-pass guess |
Two-pass AI review + LSP symbol extraction + compiler auto-fixing |
| Formatting |
Raw text output |
Auto-runs workspace formatters (Prettier, Black, gofmt, rustfmt) |
| Code Guard |
❌ None |
Proactive background analysis with non-intrusive CodeLens inline fixes |
| Data Privacy |
Sends raw code |
Local PII/credential scrubbing (HIPAA/GDPR compliant) |
| Resilience |
Fails on API 429/500 |
Automatic multi-provider Fallback Chain |
✨ Features Breakdown
🤖 Multi-Provider & Local AI
- Cloud Providers: Anthropic Claude, OpenAI (GPT-4o, o1, o3-mini), Google Gemini, OpenRouter.
- Local / Air-Gapped AI: Ollama and LM Studio (DeepSeek-Coder-v2, Qwen2.5-Coder) with custom BaseURL endpoints.
- Automatic Fallback Chain: If the primary provider hits a quota or server error, LangShift automatically retries with the next configured fallback provider.
🧪 Proactive Code Guard & Instant Health Check
- 5-Second Code Health Check: Displays a visual Maintainability Index score (0–100), auto-detected migration targets, and 1-click modernizer in the Sidebar.
- Native Inline CodeLens Fixes: Background save checks render non-intrusive
⚡ LangShift Guard: [ Preview Fix ] actions above functions with potential issues.
- Side-by-Side Diff Preview: Clicking [ Preview Fix ] opens a read-only diff editor comparing the original code with proposed fixes for 1-click approval.
🚀 Parallel Subagent Engine & Smart Caching
- Multi-Subagent Worker Execution: Parallel folder conversions process files concurrently across up to 4 worker subagent tasks.
- Parallel Chunking: Splits large files (>400 lines) into logical chunks converted in parallel.
- Two-Tier Smart Caching: Sub-millisecond in-memory L1 cache map + persistent L2 Memento LRU cache.
🔒 Enterprise Security & Compliance
- Local PII Scrubbing: Redacts emails, private IPv4 ranges, AWS keys, Bearer tokens, credit cards, and SSNs locally before API dispatch.
- Atomic File Writes: Writes output to
.langshift-tmp before swapping, ensuring zero file corruption.
- Automated Backups: Saves original code in
.langshift-backups with 1-click restore and target artifact cleanup.
⚡ Quick Start
1. Installation
Install LangShift directly from the VS Code Extensions View (Ctrl+Shift+X or Cmd+Shift+X) by searching for LangShift.
Click 🔑 Configure AI Provider & Key in the LangShift Sidebar or run:
Ctrl+Shift+P (or Cmd+Shift+P) ──▶ "LangShift: Configure API Key"
3. Convert a File
- Method A (Rename): Rename any file in the File Explorer (e.g.
Calculator.py → Calculator.java).
- Method B (Active File): Click 🔄 Convert Active File in the Sidebar.
- Method C (Folder): Click 📁 Convert Folder (Multi-Agent) to translate an entire directory.
⚙️ Configuration Settings
Every feature in LangShift is fully customizable in VS Code Settings (Ctrl+, or Cmd+, search "LangShift"):
| Setting |
Type |
Default |
Description |
langshift.aiProvider |
string |
"anthropic" |
Active AI provider (anthropic, openai, gemini, openrouter, ollama, lmstudio) |
langshift.enableCodeGuard |
boolean |
true |
Enable background Code Guard with inline CodeLens fix prompts |
langshift.enableInstantHealthCheck |
boolean |
true |
Enable instant 5-second Maintainability Index report in Sidebar |
langshift.maxConcurrentChunks |
number |
3 |
Maximum concurrent chunk conversions for large files (3x–5x speedup) |
langshift.maxSubagentWorkers |
number |
4 |
Maximum concurrent subagent tasks for batch folder conversions |
langshift.autoFormatConvertedCode |
boolean |
true |
Run workspace document formatters (Prettier, Black, gofmt) on converted files |
langshift.piiScrubbing |
boolean |
false |
Redact sensitive PII and API keys locally before prompt dispatch |
langshift.fallbackEnabled |
boolean |
false |
Auto-retry with fallback providers on API 429/5xx errors |
langshift.ollamaBaseUrl |
string |
"http://localhost:11434/v1" |
Base URL endpoint for local Ollama server |
langshift.lmstudioBaseUrl |
string |
"http://localhost:1234/v1" |
Base URL endpoint for local LM Studio server |
👥 Team Configuration (.vscode/langshift.team.json)
Project teams can commit a shared configuration file to their repository root:
{
"aiProvider": "ollama",
"ollamaModel": "deepseek-coder-v2",
"piiScrubbing": true,
"enableCodeGuard": true,
"maxConcurrentChunks": 3
}
Run LangShift: Create Team Settings Template in the Command Palette to generate this file automatically.
🛡️ License
Distributed under the Apache License 2.0. See LICENSE for more information.
Built with ❤️ for software engineering teams by Abneesh Singh.
