WebMCP Lens

WebMCP Lens gives you a clear, real-time view of every WebMCP tool exposed by a website — including its definition, schema, availability, execution results, and AI tool-selection performance. Instead of switching between code, logs, prompts, and separate testing tools, you can review the entire WebMCP implementation from one browser side panel. 1. Everything You Need to Test WebMCP 🔎 Discover and inspect tools Automatically detects all tools registered by the current page, including tools added or removed dynamically after the page loads. View: - Tool names and descriptions - Input and output schemas - Current availability - Complete tool definitions in a readable format ▶️ Invoke tools directly Test any exposed tool using: - A natural-language prompt with Try a Prompt - A generated input form - Raw JSON 📡 Live logging See WebMCP activity as it happens: - Tool registrations - Status changes - Tool-list requests and responses See the response, execution status, duration, and errors immediately. No separate test page. No digging through console output. 2. Evaluate Whether AI Can Actually Use Your Tools A valid tool is not necessarily a tool an AI model will understand correctly. Create evaluation suites to check whether an AI model can: - Understand the request - Select the correct tool - Generate the right arguments - Avoid incorrect or forbidden tools - Behave consistently across repeated runs Review tool-selection accuracy, argument matching, consistency, and invalid model responses from one dashboard. 3. Find Problems Before They Reach Production Run a one-click audit across: - Definitions - Schemas - Tool-set design - Safety WebMCP Lens highlights errors, warnings, recommendations, and passed checks, helping you quickly identify unclear descriptions, broken schemas, overlapping tools, unsafe actions, and other common implementation problems. 4. Build Repeatable Tests Save any invocation as a reusable functional test. Run individual tests or complete test suites after website updates to confirm that tools still accept the expected inputs and return the right results. It turns one-off debugging into repeatable QA. 5. Private by Design No analytics. No tracking. No external storage. All extension data stays locally on your device. The only optional network request occurs when you choose Gemini Cloud evaluations and provide your own API key. Nothing is sent to the extension developer. ---- WebMCP Lens helps you move from: “The tools are registered.” to: “The tools work, pass QA, and can be used reliably by AI agents.” ---- Built by SWC Consulting A digital analytics and data technology consultancy focused on serious measurement infrastructure.