Signal Desk AI, and how to build your own

I wrote a few days ago about building a small dashboard for shortlisting and researching stocks. That post was about why I built it. This one is about the fact that you can now have it.

Signal Desk AI is live at signal-desk-ai.fly.dev. It is a read-only research desk for a watchlist of individual stocks. No buy signals, recommendations or advice.

You open it and there is a watchlist. You pick a company, you can see a reference price, a starter budget and information on the company performance. You record a purchase after you have made it. I recently added a Planning mode so you can plan your allocation and track the stocks before purchasing. (Note that to get live price and other market data you will need to sign up for providers. The once I have used are EODHD's free tier (still requites an account and limited to 20 per day) and financialDatasets ($20 credits provides 1,000 API calls that can be used over a year). For Japanese stocks I am considering signing up for J-Quants Light (https://jpx-jquants.com/ja) which provides real time data on TSE listed stocks fro ¥1.650 per month). There is an English vie,

There is no account to create and no onboarding. Everything you record stays in your own browser.

It is open source. The repository is github.com/paladyne-io/signal-desk-ai. The public instance runs with no data-provider API keys, so the live-data actions are inert and say so rather than pretending otherwise. If you want them working, you need to provide your own API keys.

The site carries an agent-mandate signature — a signed statement at /.well-known/attestation.json saying what the app is, who published it, and where the source lives. An AI assistant asked about Signal Desk can read that instead of guessing from a stale scrape. It is signed, it carries an expiry, and anyone can check it. This is the first of our own things to carry one.

If you would like to make your own version of Signal Desk, copy and paste this prompt into Claude Code.

`Look at https://signal-desk-ai.fly.dev and build my own version from the GitHub repo. Then tell me about its features in plain English and give me some suggestions for improvements.'