WebMCP Local Agent

WebMCP Local Agent is a developer inspection and debugging tool that lets you discover, test, and run Model Context Protocol (WebMCP) tools registered on web pages alongside a local AI model. Inspect, debug, and automate tools directly in your browser: • Tool Inspector: Instantly view all WebMCP tools exposed by the active page, including names, schemas, arguments, and return types. • Live Execution & Tracing: Trigger page tools manually or test them in real time with detailed parameter inputs and structured call traces. • Local Model Side Panel: Chat with your local AI model (such as Ollama) right beside your active webpage to orchestrate tools without cloud latency or third-party API fees. • Knowledge & Graph Explorer: View entity memory, ontologies, and operational logs generated during tool runs. • Tab-Scoped Workspaces: Inspect tools on the tab in front of you without clutter or cross-tab pollution. How to use WebMCP Local Agent: 1. Ensure Ollama or your favorite local model server is running (default: http://127.0.0.1:11434). 2. Open the Chrome side panel by clicking the WebMCP extension icon in your toolbar. 3. Browse to any web page or web app implementing WebMCP tools. 4. Explore available tools in the MCP tab, or ask the side panel agent to interact with the page for you. 5. Review real-time execution traces, input parameters, and results in the Traces tab. Privacy & Local-First Architecture: WebMCP Local Agent does not require an account, API keys, or cloud subscriptions. All inference happens on your machine via your configured local model host. No page content, prompts, browsing history, or telemetry are ever sent to external cloud servers. Everything stays private on your computer. Support & Source Code: WebMCP Local Agent is open source. For documentation, troubleshooting, feedback, or feature requests, visit our GitHub repository: https://github.com/trentbrew/webmcp-ollama-agent