The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
ā¢ 1d
I was tired of 10 tabs, 10 API keys, 10 subscriptions to code with AI.
So I built TEOS AI Gateway in Alexandria, Egypt šŖš¬
WHAT IT IS:
One 11.2 MB local installer (Tauri + Rust) that gives you ONE endpoint (localhost:8082) for everything.
FEATURES:
ā 57 AI Providers: OpenAI, Claude, Gemini, Groq, NVIDIA NIM, Ollama, LM Studio +50 more
ā 11 Coding Agents Native: Claude Code, Codex, Cline, Roo Code, Aider, OpenCode...
ā Smart Auto-Fallback: if provider fails, auto-switches
ā Keys stay local: AES-256 encrypted, never leaves your machine
ā MCP Supported
ā Start FREE with NVIDIA free tier / Groq free / Ollama local
PRICING:
$39 one-time for first 50 makers (then $99) - No subscription. Lifetime.
Source included AGPL-3.0.
LINK:
š teos.elmahrosa.org
š Live on Product Hunt: producthunt.com/products/teos-ai-gateway
Built for developers who want control, privacy, and speed.
Stack: Rust, Tauri, TypeScript, MCP, Local-first
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
Where should an agentās permission check live?
I built DeadBolt at NextEleven to put that check at the point where a tool action actually runs, outside the modelās instructions. Stable 2.0.0 is now available, free and open source under Apache-2.0.
For example: an agent requests a file write. Your executor asks DeadBolt first. A denial leaves the protected function uncalled. You can set tool policies, require one-use approvals, pause a run, or kill its future admissions. Store or transport failure denies protected dispatch.
You can embed it in Rust, use a local sidecar from Python or another language, or wrap a trusted stdio MCP server.
The boundary matters: DeadBolt protects routes you connect. It is not a sandbox for arbitrary code, and kill does not automatically cancel work already running.
Install, examples and captioned walkthroughs:
https://nexteleven.dev/deadbolt
If youāre building agents, what would make a permission gate easier to integrate into your stack? Iām happy to answer setup questions and consider useful feedback.
nexteleven.dev
DeadBolt ā Give agents tools. Keep control.
A local execution gate for software agents, built by NextEleven.
For me, a small MCP example would help: approve a file write, then show what happens if the agent changes the path or content before execution. Is the one-use approval tied to those exact arguments?
A full marketing site for a concept AI product, built end to end with Claude Opus 5.5 and deployed on Vercel.
What's included
ā¢ 9 pages: home, pricing, about, contact, 2 product pages, resources, article, legal
ā¢ Animated loader, drag-to-scan hero, pinned scroll sections, page transitions
ā¢ Interactive demos: approve/reject flows, live counters, accordions, filters
ā¢ Every illustration generated by code. No stock images.
ā¢ Mobile, tablet and desktop layouts
Build stats
ā¢ 2h 21m from first prompt to live site
ā¢ $17.89 in AI usage at API prices
ā¢ ~3,970 lines of code
lintel-ai.vercel.app
Brand, copy and artwork are original.
Want a site like this for your product? Send me a message with your link and I'll tell you what I'd build.
Top-notch design.
I Built an AI-Powered Quiz Platform in 3 Day Using Only My Phone. The Odds Were (and are) Against Me.
Meet Aicai Quizzes by CFMoon, a quiz creation and exploration platform I built in just three days.
š Try it here: https://aicai.cfmoon.net
Aicai lets you generate quizzes with AI, share them with others, and automatically grade answers using AI, especially useful for open-ended and paragraph-length responses.
But the interesting part isn't just what I built. It's how I built it, and under what circumstances.
1. Building from outside the usual startup ecosystem
I'm not based in the US or another major tech hub. That means dealing with a different set of constraints, from limited access to payment infrastructure like Stripe to fewer opportunities, no access to GoFundMe, Kickstarter, or startup funding and other resources that developers in more privileged ecosystems often take for granted.
Building software is global. Access to the infrastructure surrounding it isn't always so convenient.
2. One person, one laptop, one phone, and AI
I did this entirely on my own.
No co-founder, no team of developers, no marketing department, and no established professional network to lean on. Just my experience as a software engineer, AI tools, and the hardware I already had: an aging laptop and a smartphone.
3. I built Aicai using only my phone. Yeah, why not use my computer? With the little time I have each day, it is hard to sit down and code, so I built it on the go, while on the street, connecting to public WiFi, all while running my already strenuous life.
AI didn't replace my engineering experience. It amplified what I could do with the resources available to me, and it really made the whole thing easier.
4. Almost no runway
I had very little money left to work with. This wasn't a project backed by investment or a comfortable financial cushion.
The Lovable free trial, made possible through Contra Challenges and Lovable, gave me the opportunity to build and ship the product without spending money I couldn't afford to spend.
A big thank-you to both for making that possible.
5. Building while technically homeless
There's one more detail worth mentioning: I'm currently living in a hotel and paying for accommodation by the day.
In other words, I built and shipped a software product while dealing with housing insecurity and the uncertainty that comes with it.
It's very difficult, and I wouldn't recommend it.
Why I'm proud of Aicai
The platform was built with proper software engineering practices and attention to code quality, maintainability, and reliability. Whop, a great platform, handles payments checkout, and Lovable Cloud handles users, registrations and AI features, which I hope to get enough runway to keep it online.
Three days is a short development window, but shipping something meaningful in that time takes more than prompting an AI to generate an application. It takes knowing what to build, making architectural decisions, evaluating the output, and getting the product ready for users, all while dealing with an already difficult life.
Aicai is a small product, but it's also a demonstration of what's possible when willpower beats the odds. That's where my experience as an engineer and tinkerer mattered, I could do it under pressure, under actual constrains and difficulties, and no degree will ever teach you that, only life.
No funding round. No team. No fancy setup. No rich uncle. Just an idea, the skills to execute it, and the willingness to make something with what I had.
This is what I'm building at CFMoon: useful software, built with professional expertise and quality in mind, always.
Explore Aicai: https://aicai.cfmoon.net
