🔮 Chipz AI - VS Code Extension

AI-powered Verilog RTL code generation for hardware designers. Generate synthesizable Verilog code from natural language specifications using state-of-the-art LLMs.

✨ Features

• 🤖 AI-Powered Generation: Generate Verilog code from natural language descriptions
• 🎯 Multiple Backends: Support for AWS Bedrock (Claude), Google Gemini, and local models
• 🚀 Fast & Easy: Simple interface integrated directly into VS Code
• 📝 Smart Insertion: Insert generated code directly at cursor or replace selection
• 🎨 Beautiful UI: Dark purple themed interface with syntax highlighting
• 🔌 Flexible Deployment: Works with local server or cloud API

📦 Installation

Method 1: Install from VSIX (Manual)

1. Download the latest chipz-ai-X.X.X.vsix file
2. Open VS Code
3. Open Command Palette (Cmd+Shift+P on Mac, Ctrl+Shift+P on Windows/Linux)
4. Type "Install from VSIX" and select Extensions: Install from VSIX...
5. Select the downloaded VSIX file

Or via command line:

code --install-extension chipz-ai-0.1.0.vsix

Method 2: Install from VS Code Marketplace

(Coming soon)

🏗️ Building

Local build (development)

cd vscode-extension/client
npm install
npm run package

The default API endpoint in a local build is http://localhost:8000.

Release build (for testers / production)

DEFAULT_API_URL=https://chipz-production.up.railway.app npm run package

This bakes the Railway URL into the extension so testers connect to production by default — they only need to paste their API key in settings.

---

🚀 Quick Start

1. Start the Backend Server

Option A: Local Server (for development)

# From the RTL project root
python utilities/start_server.py

The server will start at http://localhost:8000

Option B: Cloud Server (production)

The hosted server runs at https://chipz-production.up.railway.app. Use a release build of the extension (see Building above) or set the endpoint manually in VS Code settings.

2. Configure the Extension

1. Open VS Code Settings (Cmd+, or Ctrl+,)
2. Search for "Chipz AI"
3. Set Chipz-ai: Api Endpoint if needed:
   • Local dev: http://localhost:8000
   • Production: https://chipz-production.up.railway.app
4. Set Chipz-ai: Api Key to your API key (required for the hosted server)

3. Generate Your First Module

1. Open or create a Verilog file (.v or .sv)
2. Open Command Palette (Cmd+Shift+P / Ctrl+Shift+P)
3. Type "Chipz AI: Generate Verilog from Specification"
4. Enter your specification in the panel
5. Click "🚀 Generate Verilog"
6. Insert the generated code into your file!

📚 Usage Examples

Example 1: Simple Multiplexer

Specification:

Create a 2-to-1 multiplexer with inputs a, b, sel and output out.
When sel is 0, output a. When sel is 1, output b.

Generated Code:

module mux2to1 (
    input wire a,
    input wire b,
    input wire sel,
    output wire out
);

assign out = (sel) ? b : a;

endmodule

Example 2: 4-bit Counter

Specification:

Create a 4-bit up counter with:
- Clock input: clk
- Reset input: reset (active high, synchronous)
- Enable input: enable
- Output: count[3:0]

The counter increments on rising clock edge when enabled.

Generated Code:

module counter_4bit (
    input wire clk,
    input wire reset,
    input wire enable,
    output reg [3:0] count
);

always @(posedge clk) begin
    if (reset) begin
        count <= 4'b0000;
    end else if (enable) begin
        count <= count + 1'b1;
    end
end

endmodule

⚙️ Configuration

Extension Settings

Setting	Default	Description
chipz-ai.apiEndpoint	http://localhost:8000	Server endpoint URL
chipz-ai.apiKey	""	API key for authenticating to the Chipz AI backend — not your Anthropic key, see below
chipz-ai.timeout	120000	Request timeout (ms)
chipz-ai.preferredBackend	bedrock	Default backend config
chipz-ai.autoInsert	false	Auto-insert generated code
chipz-ai.showNotifications	true	Show notification messages

Backend Configurations

The server supports multiple AI backends:

• AWS Bedrock (Claude) - High quality, production-ready
• Google Gemini - Fast and cost-effective
• OpenAI
• Anthropic (direct API) - Bring Your Own Key, see below

Configure backends in the server's config files (see backend README).

🔑 Bring Your Own (Anthropic) API Key

Two Claude models in the chat model dropdown — Claude Sonnet 5 (your Anthropic key) and Claude Opus 5 (your Anthropic key) — run on your own Anthropic API key instead of our shared backend credential, so usage is billed directly to your own Anthropic account instead of ours.

This is not the same thing as a claude.ai Pro/Max subscription. You need a separate, pay-as-you-go API key from console.anthropic.com — your claude.ai chat login won't work here.

To set it up:

1. Get a key from console.anthropic.com → Settings → API Keys → Create Key.
2. In VS Code, open the Command Palette (Cmd+Shift+P / Ctrl+Shift+P) and run Chipz AI: Set Anthropic API Key.
3. Paste the key when prompted. It's validated against Anthropic (at zero token cost) before being saved.
4. The key is stored in VS Code's secure SecretStorage (OS Keychain), never in settings.json and never sent anywhere except directly with your own requests.

Other commands: Chipz AI: Test Anthropic API Key (re-validates what's stored) and Chipz AI: Clear Anthropic API Key (removes it).

If you select one of these two models without a key set (or with a key that's been revoked), the chat will show a clear error telling you to run the "Set" command — nothing is sent to Anthropic on your behalf without your own key attached.

🎨 Commands

Access via Command Palette (Cmd+Shift+P / Ctrl+Shift+P):

• Chipz AI: Generate Verilog from Specification - Open generation panel
• Chipz AI: Test Server Connection - Verify server connectivity
• Chipz AI: Open Settings - Quick access to extension settings
• Chipz AI: Set Anthropic API Key - Set your own Anthropic key for BYOK Claude models (see Bring Your Own (Anthropic) API Key)
• Chipz AI: Test Anthropic API Key - Re-validate the currently stored key
• Chipz AI: Clear Anthropic API Key - Remove the stored key

🔧 Troubleshooting

Server Not Connected

Problem: Status bar shows "❌ Not connected"

Solutions:

1. Check if server is running:

   curl http://localhost:8000/health
   

2. Start the server:

   cd vscode-extension/backend
   python start_server.py
   

3. Verify endpoint in settings matches your server URL

4. Test connection: Use Command Palette → "Chipz AI: Test Server Connection"

Generation Takes Too Long

Problem: Request times out after 2 minutes

Solutions:

1. Increase timeout in settings (chipz-ai.timeout)
2. Check AWS credentials (for Bedrock backend):

   aws sts get-caller-identity --profile chipz
   

3. Try a different backend (Gemini is typically faster)

AWS Authentication Errors

Problem: "Backend not available" with Bedrock

Solutions:

1. Refresh AWS SSO credentials:

   aws sso login --profile chipz
   

2. Verify Bedrock access:

   aws bedrock list-foundation-models --region us-east-1
   

3. Check IAM permissions include bedrock:InvokeModel

Generated Code Has Errors

Problem: Code doesn't compile or has syntax errors

Solutions:

1. Make specification more detailed - Include all ports, behaviors, edge cases
2. Specify module name and interface explicitly
3. Try regenerating - AI models can produce different results
4. Use a different backend - Some models may produce better results

📖 Documentation

• Backend Server Setup
• AWS Deployment Guide
• API Documentation

🚀 Publishing to the VS Code Marketplace

Account Details

Field	Value
Publisher ID	chipz-ai
Publisher name	Chipz AI
Microsoft account	admin@chipz-ai.com
Password hint	B+6!
Manage URL	https://marketplace.visualstudio.com/manage/publishers/chipz-ai

Steps to publish a new version

1. Bump the version in package.json (e.g. 0.1.0 → 0.1.1)

2. Publish directly via CLI:

   cd vscode-extension/client
   DEFAULT_API_URL=https://chipz-production.up.railway.app npx vsce publish -p $VSCE_PAT
   

   VSCE_PAT is saved in ~/.zshrc. Run source ~/.zshrc if it's not in your current shell. If the PAT has expired, generate a new one at https://dev.azure.com/ signed in as admin@chipz-ai.com (password hint B+6!), scope: Marketplace → Manage, expiry: 1 year.

---

🚂 Railway (Backend Deploy)

Account Details

Field	Value
Login method	GitHub OAuth — chipz-ai GitHub account
Project name	gleaming-appreciation
Project ID	25f840f3-8c48-435a-acbf-6880912a5df0
Production URL	https://chipz-production.up.railway.app
Dashboard	https://railway.app/project/25f840f3-8c48-435a-acbf-6880912a5df0

Deploys

Railway auto-deploys on every push to main when files under remote_backend/, src/, config/, or requirements files change (controlled by railway.toml watchPatterns). No manual deploy command needed — just push to main.

3. Verify — the new version appears on the Extensions tab within a few minutes: https://marketplace.visualstudio.com/manage/publishers/chipz-ai

Visibility

The extension is currently set to Public. To make it unlisted (invite-only):

• Click the extension → Details tab → change Availability to Private

---

🤝 Contributing

This is an MVP/internal tool. For bugs or feature requests, contact the Chipz AI team.

📝 License

MIT License - See LICENSE file for details

🔗 Links

• GitHub Repository: https://github.com/chipz-ai/vscode-extension
• Backend API: See backend README
• Issue Tracker: https://github.com/chipz-ai/vscode-extension/issues

💡 Tips

• Be specific - More detailed specifications produce better code
• Include edge cases - Mention reset behavior, clock edges, etc.
• Name things - Specify module names and signal names
• Test locally first - Use local server for development
• Save history - Generated code is shown in panel until closed

🎯 Roadmap

• [ ] Testbench generation
• [ ] Code validation with iverilog
• [ ] Multi-file project generation
• [ ] Code refactoring suggestions
• [ ] Integration with synthesis tools

---

Made with 🔮 by Chipz AI