Quickstart
Go from zero to receiving improvement proposals in 5 minutes.
Prerequisites
- A product or app with users (web app, DeFi protocol, AI tool, internal tool)
- 5 minutes of setup time
Create your account
Sign up at specter.draftlabs.org/signup. You'll be on the Free plan with 1 product, 500 signals/month, and 3 proposals/month.
Name your product
After logging in, the onboarding wizard asks you to name your product. This is the app or service you want Specter to monitor. Examples:
Describe what your product does
Write a short description of your product and what success looks like. This becomes a context that Specter uses to understand your signal data. Be specific:
“A token swap interface on Base. Users select tokens, enter amounts, and execute swaps. Success = completed swap with low slippage. Failure = failed transaction, abandoned swap, or user rage-clicking the swap button.”
Choose your integration method
Pick the integration that fits your stack. The fastest option is the script tag — paste one line of HTML and you're done.
Script Tag (no-code)
Auto-tracks page views, JS errors, rage clicks, and form submissions.
<script
src="https://specter.draftlabs.org/s.js"
data-api-key="YOUR_API_KEY"
async
></script>Webhook URL (Zapier / Make.com)
Point any Zapier or Make.com workflow at your webhook URL:
https://specter.draftlabs.org/api/ingest?key=YOUR_API_KEYSDK (npm)
Full control. Install the SDK and track custom events:
npm install @specterorg/sdkimport { Specter } from "@specterorg/sdk";
const specter = new Specter({
apiKey: "YOUR_API_KEY",
});
// Track any user action
specter.track("swap_completed", {
token_in: "ETH",
token_out: "USDC",
amount: 1.5,
});Send a test signal
Send a quick test signal from your terminal to verify everything is connected:
curl -X POST https://specter.draftlabs.org/api/signals \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_API_KEY" \
-d '{"type": "thumbs_up", "payload": {"page": "home"}}'You should get back: { "received": 1 }
What happens next?
Every 6 hours — Specter's analysis engine scans your signals from the last 24 hours. If there are 5+ signals, it runs AI analysis and generates proposals.
New proposals — Appear on your Dashboard as “Pending”. You can approve, reject, or defer each one.
When you approve — A proposal and implement the change, Specter automatically tracks before/after signal metrics to verify the improvement worked.
Can’t wait 6 hours? — Use Analyze Now to run on-demand analysis for $1 per run.