Inspect, run and test WebMCP tools

A Chrome side panel for developers building WebMCP tools. See every tool a page registers, run them by hand, and let an AI agent use them with the model and provider of your choice.

Add to Chrome Try the demo page

Needs Chrome 155 or later with WebMCP turned on: open chrome://flags/#enable-webmcp-testing, set it to Enabled and relaunch Chrome.

What it does

Inspect Tools in the page and in every iframe, with their schemas, annotations and origin.
Lint WebMCP rules, description quality, and instruction-like or hidden text.
Run From a form generated from the schema or from raw JSON, with a history of runs.
Agent Chat with OpenAI, Claude, Gemini, OpenRouter, a local model or Chrome's built-in AI.
Voice Talk to the agent through Gemini Live. Off by default.
Override Try a different description or schema without redeploying. The page is never changed.
Changes and logs How tools change over time, a timeline of events, and every model call.
Export Tools as JSON, MCP, OpenAI, Anthropic or Gemini declarations, and evals.

Your data

No account, no analytics, and no servers of ours. Toolbench runs in a page only while its panel is open on that tab. If you use the agent, your messages, the page's tools and their results go to the AI provider you choose, and only there. API keys stay in your browser. Read the privacy policy.