[ Roadmap ]
Roadmap
Right now, you can test your MCP application manually, record sessions, and automate them in 5 minutes.
Shipped
-
Manual test: tool calls
Call any MCP tool with custom arguments and inspect the result.
-
Manual test: resources
Read any resource URI and inspect the response.
-
Widget resource
Preview and interact with widget resources. Inject mock data, switch platform and viewport.
-
Automation E2E Testing
Record a session and replay it as a repeatable test.
-
Assertions
Per-point assertion modes: exact, shape, flaky, ignore.
-
Testing with ChatGPT, Claude, Copilot
Expose your local server on a public tunnel URL and connect it to any AI that supports MCP.
-
Tags and run all
Organize tests with tags. Run all or filter by tag.
Incoming
-
Better assertions
[Description of planned assertion improvements.]
-
AI-native use
Let Claude or another AI call your MCP application directly through the studio.
-
Programmable logic
Write custom code to handle special cases in your tests.
-
CI/CD integration
Run your test suite in a pipeline with a single command.
Request a feature
Have something in mind? Reach out and tell us what you want to do and why it matters.