A VS Code extension that adds AI-powered helpers to your editor:
- Generate Unit Tests — produces idiomatic tests for the active file/selection.
- Generate Design Document — high-level design doc in Markdown.
- Code Review — structured review (bugs, security/OWASP, perf, style, tests).
- Generate LLD — low-level design with mermaid class & sequence diagrams.
- Review PR — review a
git diffagainst any base branch (e.g.origin/main). - Sidebar Chat — conversational assistant in the Activity Bar.
- MCP Servers — connect to Model Context Protocol servers via stdio.
Quick start
npm install
npm run build
Press F5 in VS Code to launch the Extension Development Host.
Configure
Open Settings → search for aiDevTool:
| Setting | Description |
|---|---|
aiDevTool.provider |
openai | anthropic | azure-openai | ollama |
aiDevTool.model |
Model name (e.g. gpt-4o-mini, claude-3-5-sonnet-latest, llama3.1) |
aiDevTool.apiKey |
API key (will be migrated into SecretStorage on first prompt) |
aiDevTool.endpoint |
Custom endpoint (Azure OpenAI / self-hosted) |
aiDevTool.mcpServers |
Array of { name, command, args } for stdio MCP servers |
Example MCP config:
"aiDevTool.mcpServers": [
{ "name": "filesystem", "command": "npx", "args": ["-y", "@modelcontextprotocol/server-filesystem", "${workspaceFolder}"] }
]
Commands
All available via the Command Palette (AI: prefix) and the sidebar toolbar:
AI: Generate Unit TestsAI: Generate Design DocumentAI: Review Current FileAI: Generate Low-Level DesignAI: Review Pull RequestAI: Connect to MCP Server
Project Layout
src/
extension.ts # activation
ai/aiService.ts # provider-agnostic LLM client
features/ # one file per feature
mcp/mcpManager.ts # stdio MCP client
webview/ # sidebar webview
util/editor.ts # editor helpers
Security
- API keys are stored in VS Code SecretStorage after first entry.
- Webview uses a strict CSP with a per-load nonce.
- The PR review feature shells out to
git; ensure your repo is trusted.