Loading

Kagura AI | Product Hunt
Now with MCP support for Claude Code

Let your AI agent
test what it builds

AI writes the code. But who checks if it actually works? Now your agent can open a browser, click through flows, and verify before shipping.

View on GitHub
$ npm install -g @kagura-run/cli
Scroll to explore
0
Frame-perfect control
0
Command to start
0
% Open source
0
AI credits used

Everything your agent needs

Browser control, email handling, and CI/CD — all via CLI or MCP.

🌐

Browser Control

Navigate, click, fill, screenshot — full Playwright power via simple commands.

📧

Email Skills

Handle magic links, OTPs, and email verification. No more "check your email" blockers.

🔍

DOM Analysis

Structured accessibility tree so your agent understands the page, not just sees it.

📼

Action Recording

Every action logged. Replay in CI/CD without rewriting tests.

MCP Native

First-class Claude Code integration. Just add to your MCP config.

🔌

HTTP API

Works with any agent. Codex, Cursor, or your own — if it can curl, it can test.

Ready to get started?

Copy this prompt and give it to your AI agent. It'll handle the rest.

Agent Prompt

Works with Claude Code, Codex, Cursor, or any coding agent.