Web testing
integration for Claude Code
Create extensive web-based tests with AI agents and Beryl: self-healing, intent-based, and ready to run in Beryl's cloud, on a schedule, or in your CI pipeline.
One command signs you in and wires the Beryl MCP into your AI agent: Claude Code, Cursor, Copilot, or opencode. Nothing else to configure.
run this in your terminal ↓
Create tests once. Run them everywhere.
Create extensive web-based tests from natural language using AI agents and Beryl, then run the same tests in Beryl's cloud, on a schedule, or in your CI pipeline.
Describe the flow in your AI agent. Beryl turns it into an extensive web test.
Create a test once and run the same test across preview, staging, and production.
Whenever something breaks, you get alerted on Slack, not by a user.
Instruction blocks set up state before a test and clean up after it, so even tests that change the application are safe.
Watch it work in Claude Code
Watch the replay
Every step of the run, on beryl.so.
Aurora Lamp
$49.00
Sol Sconce
$29.00
Dune Shade
$35.00
Teardown
What you get with Beryl
Created with AI agents
Describe the flow in plain language and get an extensive web test.
Self-healing, intent-based
Intent, not selectors: drift heals, real breaks fail loudly.
Scheduled automatic runs
Put the suite on a schedule and Beryl runs it for you.
One test, every environment
No environment config juggling: create once, run on preview, staging, and production.
Slack alerts
When something breaks, the team hears it on Slack right away.
Variables & file secrets
Set once, use across every test, nothing hardcoded.
Ask for tests like you ask for code
One command in your repo, then it's part of every AI agent session. Free while in beta.