WebMCP Active — Polyfill + Claude API
WebMCP Integration
This portfolio is a live MCP server. It exposes 5 structured tools via the W3C WebMCP API — powered by GPT-4o-mini for intelligent Q&A. AI agents, browser assistants, and assistive technologies can query projects, search skills, and interact with this site programmatically.
How it works
🌐
Portfolio loads
Remix SSR + React hydration
→
⚙️
Tools registered
navigator.modelContext.provideContext()
→
🤖
Agent calls tool
ask_about_shiva(question)
→
🧠
Claude responds
Portfolio context + Claude Sonnet
Portfolio AI — powered by GPT-4o-mini via WebMCP
0 messagesHey! I'm the AI assistant powering Shiva's portfolio. This conversation is powered by GPT-4o-mini via the WebMCP `ask_about_shiva` tool. Ask me anything about Shiva's projects, skills, or experience!
Integration Code
WebMCP tools registered via the W3C spec, with GPT-4o-mini as the AI backend.
webmcp-tools.jsJavaScript
// Register tools via W3C WebMCP API
// Spec: https://webmachinelearning.github.io/webmcp/
navigator.modelContext.provideContext({
tools: [
{
name: 'ask_about_shiva',
description: 'Ask any question about Shiva Teja...',
inputSchema: {
type: 'object',
properties: {
question: { type: 'string' }
},
required: ['question']
},
async execute({ question }) {
// Calls GPT-4o-mini with portfolio context
const res = await fetch('/api/webmcp-chat', {
method: 'POST',
body: JSON.stringify({ question })
});
const { answer } = await res.json();
return {
content: [{ type: 'text', text: answer }]
};
},
annotations: { readOnlyHint: true }
},
// ... filter_projects, get_skills, etc.
]
});