Run PartyRock Locally

✨ App Planner — PartyRock Slicing

Fill in your PartyRock app details below to generate a ready-to-paste Kiro prompt for local development.

App Info

API Endpoint

Widgets (from PartyRock edit panels)

Extra Notes (optional)

Generated Kiro Prompt

Fill in the fields above...

Step 1 — Generate index.html

Your app calls Bedrock directly using an API key. No Lambda or API Gateway needed.

Your API Details

Paste This Into Kiro

Build me a single index.html that calls Amazon Bedrock directly.

APP: [Your App Name]
DESCRIPTION: [What it does]

API:
- Endpoint: https://bedrock-runtime.ap-southeast-5.amazonaws.com/model/global.anthropic.claude-haiku-4-5-20251001-v1:0/converse
- Auth: Authorization: Bearer [YOUR_API_KEY]
- Content-Type: application/json
- Body format (Bedrock Converse API):
  {
    "messages": [{"role": "user", "content": [{"text": "..."}]}],
    "system": [{"text": "system prompt here"}],
    "inferenceConfig": {"maxTokens": 2048, "temperature": 0.7}
  }

WIDGETS:
- [List your input widgets: name, type, placeholder]
- [List your AI output widgets: name, system prompt, which inputs they use]

REQUIREMENTS:
1. Single index.html — no build tools, no server needed
2. All CSS and JS inline
3. Each AI widget calls the Bedrock endpoint with its own system prompt
4. Build the user message by combining input widget values
5. "Run All" button fires all AI calls simultaneously
6. Loading spinners per output, error messages if call fails
7. Mobile-responsive, dark theme
8. Markdown rendering via marked.js CDN
9. Copy button on each output, Reset button
10. Works by double-clicking the file (file:// protocol)

IMPORTANT:
- Only model available: global.anthropic.claude-haiku-4-5-20251001-v1:0
- Use the Converse API format (not InvokeModel)
- API key goes in Authorization: Bearer header

NO server-side code. Everything runs in the browser.

Drag your PartyRock screenshots into the same Kiro message → send.

Save

Save the generated index.html anywhere on your computer. That's the only file you need.

Step 2 — Run

Double-click index.html to open it in your browser. That's it.

Fill in the inputs → click Run All → AI responds.

If Streaming Doesn't Work on file://

Some browsers block streaming fetch on file:// protocol. Quick fix — use any tiny server:

# Pick one (all do the same thing):
npx serve .                      # if you have Node
python3 -m http.server 5000      # if you have Python
php -S localhost:5000             # if you have PHP

Then open http://localhost:5000

Troubleshooting

Making Changes

Open index.html in any text editor → edit → save → refresh browser.

Or share the file with Kiro and ask: "Change the background to dark blue" / "Add a reset button"

Sharing