Glance FAQs

Question 1

What is Glance and how does it benefit AI agents?

Accepted Answer

Glance is an MCP server that provides AI agents, particularly Claude Code, with a real browser environment. This allows AI agents to actually 'see' (via screenshots), interact with, and test web applications, moving beyond mere guesswork to actual programmatic interaction and validation.

Question 2

Is Glance configurable for different security and testing needs?

Accepted Answer

Absolutely. Glance is highly configurable via environment variables, allowing users to set browser headless mode, viewport dimensions, timeouts, and crucial security profiles (local-dev, restricted, open) that control URL access, JavaScript execution, and rate limiting.

Question 3

What types of web application testing does Glance support?

Accepted Answer

Glance supports comprehensive End-to-End (E2E) testing, visual regression (pixel-level comparison), UI interaction testing, network & console monitoring, and multi-step test scenario execution. It includes 12 assertion types for robust validation.

Question 4

What are the key features for controlling the browser and testing?

Accepted Answer

Glance offers 30 MCP tools for complete browser control, including navigation, clicks, typing, and screenshots. For testing, it provides inline screenshots, visual regression, a JSON-based test scenario runner, 12 assertion types, network/console monitoring, and session recording.

Question 5

Can Glance integrate with existing AI workflows or tools?

Accepted Answer

Yes, Glance is designed to integrate seamlessly with Claude Code via its MCP server interface. It ships with a dedicated Claude Code agent (`e2e-tester`) and a quick test runner skill (`/glance-test`), enabling AI to directly command browser actions and interpret results.

Glance

Glance

主な機能

ユースケース

主な機能

ユースケース