I just read through the site and your GitHub commits and you have done ALOT - good work AND it would be interesting to know your Cursor workflow.
p.s, i also love convex i think its amazing
What I Built
The LLM Visibility Lab — A chemistry-themed website for AI SEO and LLM visibility optimization. This project contains a landing page for my LLM Visibility + SEO agency and guides on SEO fundamentals and LLM visibility, news articles, an interactive Google Trends dashboard, and service pages. Screenshot
Live URL: https://www.llmvlab.com
How Cursor Helped
Used Cursor to set up the full-stack setup: React Router 7 with Cloudflare Workers, Contentful integration, 15+ reusable components, SEO optimization (meta tags, structured data, sitemaps), API routes for trends data, content generation tools, and a chemistry-themed design system.
Thanks! My workflow is to use Claude 4.5 Opus in MAX mode for 99% of all requests, sonnet 4.5 if I need extra context window, like when modifying gigantic documentation files. I tend to as the agent to build a 2000+ word deep research prompt for chat gpt 5.2 Pro DR agent anytime I need to pull in lots of recent context for external tools and information. I heavily use the Plan mode to go back and forth on outlining the perfect new feature to build, then let it execute on that plan autonomously, followed by heavy unit, integration, and e2e tests for whatever was built to ensure function. Usually, those tests find bugs in the implementation. I allow bugbot to run after every commit and I commit often.
Cost: infinity
Workflow productivity: super high!
I had some time off at the end of the year and I vibe coded a website to track news stories: https://frontpagestats.live
I love Cursor! Shipped in 2025 (2nd half, mostly):
- AniMoment (Web) - Take a photo and receive a gentle, spoken animist reading of place (a tiny ritual of noticing what’s actually around you).
- Decomposer (Web) - Turn messy goals into clear, executable plans with honest effort routing (You / AI / Expert) and copy-ready sub-prompts you can run step by step.
- BuddhaUR (iOS | Android) - Learn early Buddhism through conversation and get grounded guidance with tappable citations to primary sources.
- Agentic.ai (Web) - AI-powered semantic alerts that dedupe noisy coverage into story clusters, then explain what changed and why it matters (with citations).
- BloomEDU (iOS) - A focused learning app for mastering wildflowers with photo-rich lessons, habitat/elevation filters, and bite-sized quizzes that actually stick.
- Cursor Environment Docs (GitHub) - Auto-updating environment documentation that keeps Cursor smart about your project setup.
- QuickBite (Web) - Fast, fair group voting that turns a 30-minute “where should we eat?” debate into a 2-minute decision with transparent consensus scoring.
More to come!
Links to all projects here:
I’m a DJ lol and used Cursor to make a cross platform tag editor designed for DJ’s to organise their music libraries with fast, keyword-style tagging designed to be used in smart playlists. I also made the website for it … https://tagami.au/ - Download for free or use instant browser preview.
I’m also the Product Manager at a financial services company and after my experience with Cursor in 2025 I’ve convinced the firm to get Cursor for our whole dev team.
I shipped a small, private personal finance dashboard using React and Python this year. I cannot share the link or a demo, but I am proud of the custom reporting features. Cursor helped tremendously by generating the complex boilerplate needed for the Python API endpoints and accurately debugging tricky authentication scopes in the frontend.
Hope this helps!
1 Like
What I built:
My biggest project. Real-time Crypto Token Monitor & Alerting System with Admin Dashboard.
Live: Telegram: Launch @olfun_analytics_bot
Stack: Python, Redis, WebSocket, Docker, Grafana.
How Cursor helped:
-
Velocity: Moved from “Idea” to “Production” in hours, not weeks.
-
Architecture: Implemented complex patterns (Dual writing to Postgres/Files, Circuit Breakers) without getting stuck in syntax.
-
Quality Assurance: I used Cursor to generate 370+ tests covering Race Conditions, Redis Failures, and network timeouts.
More:
My full portfolio (also built entirely with Cursor):
https://paradoxlab.dev/
1 Like
Hey Josh,
i understand that it can be necessary for users to cleary see how exactly the outcome has been reached in trades. I would love to opensource that project at the end but i’m not really shure if i should do that - it’s gonna be a pretty powerful tool as you might understand already. 
I haven’t planned to give users the ability to input their own algorhythms since the internal engine is calculating everything by itself to a maximum win-rate. So there’s not really the need to give that option.
Also i do use AI for the work and there’s some machine learning in place, yes - but i won’t (yet) let models make trading decisions.
There are a few clear and unavoidable guidelines built in which losses or low profits are virtually impossible.
I’m always open and interested to hear how others use AI tools to create mechanisms to genereate profits. Is there a way to continue this discussion privately, would you like to do that?
Regards,
Smali
1 Like
After I have my algorithms programmed in, I’m going to experiment with letting agents paper trade.
joshhh, I see you are planning to let agents paper trade after implementing the algorithms. Have you considered using Redis to simulate trades and test the performance of your system without affecting real market data?
