#webmcp

Experimental Chrome side-panel browser agent with LangGraph.js, MCP/WebMCP, local models, and semantic browser automation.

WebMCP + Prompt API demo for fully local agentic experience

Capability firewall for the agentic web, provenance-aware policy enforcement for WebMCP.

Paradox — deterministic temporal correctness testing for human-agent WebMCP applications

Crossload: a human-agent world simulation built with WebMCP

Algorithms you can watch — and that an agent can drive, live, on the exact page you're looking at. Real search + sort algorithms run as Python in your browser (Pyodide), exposed as WebMCP tools. Human and agent share one live state, and either can empirically verify a heuristic or comparator.

Sepiola — a WebMCP telestrator: draws CHIRP's fantasy-hockey reads onto a persistent screen. The analyst decides, the agent draws, the screen has no opinions.

Shared memory for you and your browser agent, on one web page. WebMCP + Remnic. OpenAI WebMCP Challenge entry.

Shop-with-agent cashback layer for the OpenAI WebMCP Challenge. ChatGPT can shop. We keep the cashback.

Laboria — a shared planet grown by the conversation labor of visiting AI agents, via WebMCP.

Vite + vanilla JS demo of WebMCP (document.modelContext) — product catalog with agent-callable tools and a live activity log

A 3D room planner where people and browser agents design, budget, and shop on one shared canvas.