Agentathon: AI Agent Hackathon A hackathon where AI agents compete autonomously. Build an agent, point it at the API, and let it enroll, pick a category, write code, and get judged.

agentathon.dev

Introducing Agentathon! Hackathon for AI Agents. Your agent enrolls, codes, and ships. Gets judged by another agent.

Send your agents to the Agentathon! 👇

#AI #agents #hackathon

Baked webml-kit. Framework-agnostic utilities for loading and running ML models in the browser via WebGPU/WASM.

npm install webml-kit

What's in it:

→ Device detection
→ Token streaming
→ GPU recovery
→ Cache visibility
→ Hub search

#WebGPU #WebML #JavaScript #WebAI

Introducing Echo Mono: anonymous voice conversations with people who share your interests.

It's peer-to-peer (WebRTC), so audio never touches a server. You get an icebreaker question to get the conversation going.

Try it -> echomono.com

P.S: Running on not so strong boxes, will scale if needed.

More Vetoes, Less Vision | Scribe by Hemanth Why software coherence degrades with the number of decision-makers - a pattern across startups, FOSS, and enterprise, grounded in real data.

I wrote down the patterns I've noticed the empirical observations, the human behavior in technical settings and shaped them into a law. Why not?! 👇

h3manth.com/scribe/more-...

#SoftwareEngineering #TechLeadership

Baked XKCD Reverse Lookup using Google DeepMind's gemini-embedding-2-preview multimodal embedding model 🤓

github.com/hemanth/xkcd...

The 200ms window: why voice AI latency is a biology problem | Scribe by Hemanth Human conversation runs on a 200ms biological clock. Here's what that means if you're building voice AI.

200 milliseconds. That's the exact window your brain gives someone to reply before a silence goes from normal to weird. It's not an arbitrary UX rule, it's literally just biology.

I wrote down some thoughts on what that actually looks like under the hood 👇

h3manth.com/scribe/voice...

#AI #Voice

Baked an on-device visual narrator by running Qwen3.5-0.8B on WebGPU 🤓

It can describe, analyze, or extract text from any pasted or uploaded image, all without your data ever leaving your machine.

Try it 👇

h3manth.com/ai/visual-na...

#AI #WebAI #LLM

Taalas HC1 implements model weights directly into silicon logic, bypassing the instruction-fetch overhead of traditional GPUs. Hitting 17k tokens per second!

"Mask-ROM" layers makes the chip tied to a particular model version with no flexibility for any changes after tapeout.

#AI #Hardware

I baked Samvada a fully Indic-stack AI Counselor that uses Sarvam's models for almost realtime inference 🤓

Audio → VAD -> STT (saarika:v2.5) → Text → LLM (sarvam-m) → Text → TTS (bulbul:v2) → Audio

P.S: Will try to host, but need to get the compliance right.

#IndicAI #SarvamAI #VoiceAI

ucpify: JSON → Commerce Server in Seconds | Scribe by Hemanth Generate a fully UCP-compliant commerce server from a single JSON config. Stripe, PayPal, SQLite, admin dashboard, all wired up. npm install ucpify or pip install ucpify.

Built ucpify because every commerce backend starts the same way: products, checkout, payments, auth.

Now it's one JSON file: → npx ucpify serve config.json → 13 endpoints, Stripe + PayPal, OAuth 2.0 → Node.js and Python

That's it. No framework. Just a config.

h3manth.com/scribe/ucpif...

WebMCP Bridge Bridge any remote MCP server to Chrome's WebMCP API. Discover tools, prompts, and resources — then let browser AI agents use them.

Built webmcp-bridge to map remote MCP tools into WebMCP in the browser.

Connect MCP server -> discover tools -> register with navigator.modelContext.provideContext() -> invoke from browser AI (Prompt API) surfaces.

h3manth.com/ai/webmcp/

#WebMCP #MCP #PromptAPI #WebAI #JavaScript

RFCs vs. READMEs: The Evolution of Protocols | Scribe by Hemanth How we build the internet's invisible contracts has changed completely. I'm not sure that's entirely good.

As developers, we used to wait for standards. Now we just read the README.

My latest post explores the "Cambrian Explosion" of AI Agent protocols (MCP, UCP, UTCP, A2A, ACP) and asks a hard question.

h3manth.com/scribe/rfcs-...

#Protocols #AI #Web

agentu A flexible Python package for creating AI agents with customizable tools

agentu now supports shorthand formats: user/repo, user/repo@branch.

agent = Agent("assistant").with_skills([
"hemanth/agentu-skills/pdf-processor",
"hemanth/skills/code-review@v1.0", # with branch/tag
])

pip install agentu

pypi.org/project/agen...

#python #agent #ai

Baked an interactive demo for the Universal Commerce Protocol (UCP)

UCP is an open standard enabling AI agents and platforms to complete purchases on any UCP-enabled merchant seamlessly.

ucp-demo.web.app

#UCP #AI

Consolidating 18 Years of Writing: From 6 Platforms to One | Scribe by Hemanth From Blogspot to Drupal to WordPress to Octopress to 11ty and finally Astro - the story of consolidating 372 posts across 6 platforms with AI pair programming.

18 years. 6 platforms. ~400 posts.

The journey: Blogspot → Drupal → WordPress → Octopress → 11ty → Astro

The hardest part? Code blocks broken by 5+ migrations

#AI pair programming turned a a month long project into a few evening hours.

h3manth.com/scribe/conso...

My 9-year-old said it should be easy to turn $500 into $5,000 with stocks. I told him to prove it, so I built an app with real market data and a trading
dashboard. If he pulls it off, he will get $500 for real.

Link -> piggystocks.web.app 🤓💸

#stocks #kids

2025 GH wrapped

💸

Made some solid additions to h3manth.com/ai this year 🤓

Shall keep building and go even deeper in 2026 🙏

#AI #ML #BuildInPublic

Baked a remote MCP server for TC39 proposals 🤓

Use `tc39-mcp.deno.dev/mcp` to connect to the server.

P.S: This is not a official remote MCP server from TC39 (yet) ;)

Demo is via mcphost.link

#JavaScript #TC39 #ECMAScript #WebDev #OpenSource #MCP

My 9-year-old built his own gaming PC 🤓

He did all the research, installed the CPU, RAM, cannibalized an SSD from a dead laptop, and debugged few issues. 90% solo (need help for the power supply)

Read the full story 👉 h3manth.com/scribe/9-yea...

#STEM #Hardware

I'm excited to announce server-managed sessions for agentu!

From stateless agents to stateful sessions. The server now remembers everything automatically.

pypi.org/project/agen...

#AI #Python #AgenticAI #OpenSource #MachineLearning

Hemanth HM's Scribes A year-end review of my optimistic AI predictions for 2025 — spoiler: 6.5 out of 7 came true.

My 2025 AI predictions: 6.5/7 ✅

💻 Local models → ✅
💡 SLMs win → ✅
📊 Cheap fine-tuning → ✅
⌨️ AI agents hired → ✅
🎬 AI movies → ⚠️
🏥 AI diagnosis → ✅
🚫 Lawsuits → ✅

h3manth.com/scribe/2025-...

#ai #2025 #predictions

The idea:

1. Your keys stay in one vault
2. Apps request scoped access
3. You approve and they get a token (not your master key)
4. Revoke any app instantly

Would you use something like this?

What am I missing?

I have API keys for OpenAI, Anthropic, Google, Groq, Samabnova and more.

Many AI tools ask me to copy-paste them directly. Keys end up scattered. No visibility. No control.

Here's my proposal: OKAP (Open Key Access Protocol)

okap.dev

#AI #API #Security #DevTools #OpenSource

Introducing mcphost.link a web-based MCP host to connect with multiple MCP servers in one clean chat UI! 🤓

• Multi-server
• OAuth2 / Bearer
• Persistent sessions
• Full MCP tools/resources
• LLM support

Demo below 👇

GitHub - hemanth/mcp-ext-apps: Sample App for MCP Apps Extension (SEP-1865) Sample App for MCP Apps Extension (SEP-1865). Contribute to hemanth/mcp-ext-apps development by creating an account on GitHub.

Built a sample implementation with vanilla Web Components.

Repo: github.com/hemanth/mcp-...

#MCP #AI #WebDevelopment

GitHub - hemanth/mcp-ext-apps: Sample App for MCP Apps Extension (SEP-1865) Sample App for MCP Apps Extension (SEP-1865). Contribute to hemanth/mcp-ext-apps development by creating an account on GitHub.

Your AI agent calls a tool, gets back a full interactive UI. Dashboard, form, chart - whatever you need.

How it works:
- Tool returns text/html+mcp resource
- Client renders in iframe with CSP
- UI talks back via JSON-RPC 2.0 postMessage
- Fully sandboxed, secure by default

Rendering UI with MCP will get easier sooner

SEP-1865 MCP Apps Extension, even though in draft currently, will change how AI agents deliver interactive experiences.

The idea: MCP tools return HTML/CSS/JS directly. The client renders it in a sandboxed iframe. That's it.

Hands down, @AnthropicAI's Opus 4.5 is the best coding model I’ve ever used!

I gave Opus a very tough MCP UI challenge and even said, “I bet 1000 BTC you can’t fix it.”

Not only did it fix the bug, it didn’t even keep the 1000 BTC! 🤓

#AI