---
title: "How teams scale Playwright in CI. For humans and agents."
description: "Debug failures, catch flaky tests, and speed up tests in CI. Turn your test suite into the verification layer that lets you ship at agentic speed."
canonical: "https://currents.dev/"
---

# How teams scale Playwright in CI. For humans and agents.

Debug failures, catch flaky tests, and speed up tests in CI. Turn your test suite into the verification layer that lets you ship at agentic speed.

- [Start a 14-day free trial](https://app.currents.dev/signup?ref=hero)
- [Get a demo](https://calendly.com/d/3zq-45h-wwd/discovery-call)

## From failing to fixed, backed by evidence

- **Test evidence**: every attempt keeps its Playwright trace, screenshots, video, visual diffs, console and network logs next to the error. Open any trace in one click, without digging through CI artifacts.
- **Agent evidence**: when no test covers a change, the agent records the browser before and after its fix and shares both as one link. Reviewers see the bug, then see it gone, on the pull request and without a Currents account.
- **Flaky test detection**: flags tests that pass on retry and tracks each test's flakiness rate and history across branches, so people and agents can tell a flake from a real failure.
- **The context your agent needs**: agents fix what they can see. Currents gives yours the error details, failure history and traces over MCP, so it goes after the cause, not the symptom.
- **One click to your agent**: send any failure to Claude Code, Cursor or Codex with full context, from the dashboard, a GitHub comment, a Slack alert or your editor. Or share a link instead. People and agents open it without a Currents login.
- **Test ownership**: mark each test's owner or tag it by team, and failures mention the right people in Slack. Create or link a Jira or Linear issue from any failure, with its context attached. Agents can file the issue too, over MCP.

## Test observability that keeps up with your agents

- **Analytics & reports**: charts track run status, duration, flakiness and suite size over time, filtered by branch, tag or environment. Scheduled reports send the top regressions to email or Slack.
- **Failure classification**: AI classifies every error by category, action and target, then groups them across runs. See which locator or endpoint causes the most failures.
- **Quarantine rules**: quarantine, skip or tag tests by rule, no code change. Quarantined tests still run but can't fail the build. Rules can expire, and agents can create them over MCP.
- **Pull Request checks**: each run posts a status check and a results comment to its GitHub pull request. Make the check required, and failing tests block the merge. Also available for GitLab/BitBucket.
- **Slack alerts**: failed and flaky tests post to the Slack channels you choose, as they happen. A recovery message posts when the pipeline is green again.
- **Test and spec explorer**: rank every test and spec by failure rate, flakiness and duration, with trends over time. Agents can query the same data over MCP.

## Don't let tests become your CI bottleneck

- **Smart orchestration**: assigns spec files to CI machines from a live queue, ordered by historical duration, so every machine finishes at about the same time. No manual splitting or file renaming.
- **Fail fast and risky tests first**: runs the specs most likely to fail first, ranked by failure rate, and cancels the run on every machine after the first failure, or a threshold you set.
- **Rerun only failed tests**: rerun only the tests that failed, not the whole suite, spread across every CI machine. Works with sharded and orchestrated runs.
- **Spot instance support**: run tests on cheaper spot instances. If one is reclaimed mid-run, its spec files move to a healthy machine and the run completes. Cuts CI costs by up to 70%.
- **Fully parallel mode**: split a spec file's tests across every worker with fullyParallel. Currents still collects every result.
- **Crash-resilient streaming**: results stream in step by step as tests run. If a CI machine crashes, nothing it reported is lost.

## From your editor to CI, fully integrated

- **MCP server**: give AI agents full access to test history, flakiness, and execution context.
- **IDE extension**: ask AI to fix failures in your editor, with full test history as context.
- **GitHub, GitLab, Bitbucket**: PR status checks, AI failure summaries, and merge blocking on test failures.
- **Slack / MS Teams**: notifications for failures, flaky tests, and run results in the channels your team already uses.
- **Issue tracker**: link or create Jira/Linear issues from test failures, with visibility into existing tickets.
- **REST API & Webhooks**: stream data, trigger workflows, or build custom reports.

"The team was fascinated when they started to use Currents. They were getting real-time results, and our international communication channel was flooded with excitement." Rasa Mazutiene, Director of Engineering at Juvare.

## Why teams choose Currents

- **The memory your agents don't have**: an agent starts every session from zero. Currents keeps months of runs, flakes and failures, and hands that history to whoever picks up the fix, person or agent.
- **Red means broken. Green means ship.** People learn to ignore a flaky test. Agents can't. Currents separates flakes from real failures, so agents stop chasing noise and your team stops writing off real bugs as flakes.
- **CI in half the time, for a fraction of the bill**: every agent pull request is another CI run. Currents cuts each one's time by up to 50% and its cost by up to 70%.
- **Say yes to faster releases**: agents ship more code than anyone can review by hand. Currents proves each change works, with before-and-after evidence on the pull request and checks that block a failing merge.

## Real results from real test suites

- [n8n](https://currents.dev/posts/currents-and-n8n): 90% test flakiness reduction after adopting Currents.
- [FundGuard](https://currents.dev/posts/currents-and-fundguard): 70% CI cost reduction from running on spot instances with Currents.
- [Kittl](https://currents.dev/posts/currents-and-kittl): 50% faster test execution with Currents Orchestration for Playwright.
- [Ramp Network](https://currents.dev/posts/currents-and-ramp): 62% smaller CI fleet after Ramp moved to Playwright with Currents.
- [Sherpany](https://currents.dev/posts/currents-and-sherpany): 39% shorter test suite runtime with Currents, from 18 to 11 minutes.
- [Hover](https://currents.dev/posts/currents-and-hover): 80% testing cost reduction after migrating to Playwright with Currents.

## FAQs

### What is Currents?

Currents is a Playwright reporter, dashboard and test orchestrator for teams running Playwright tests in CI. It records every test run with its errors, traces, screenshots and videos, detects flaky tests, tracks test health over time, and speeds up CI by balancing tests across machines. AI agents use the same data through the Currents MCP server to fix failing tests.

### Which test frameworks does Currents support?

Currents is built for Playwright. It also reports results from Jest, Vitest, Cypress, the Node.js test runner and [any framework that writes JUnit XML](https://docs.currents.dev/getting-started/other-frameworks), so one dashboard can cover all your test suites.

### Does Currents run my tests?

No. Your tests run in your own CI, as they do today. The Currents reporter sends each result to the dashboard as it happens, and Currents orchestration tells each CI machine which spec file to run next.

### Why does testing matter more when AI agents write code?

Agents write code faster than anyone can review it by hand, so the test suite becomes how you know a change works. Currents makes that suite a verification layer: every run is recorded with its evidence, flaky tests are separated from real failures, and agents read the results to fix what they broke and prove the fix on the pull request.

### How do AI agents use Currents?

Through the [Currents MCP server](https://docs.currents.dev/ai/connect). Claude Code, Cursor, Codex, ChatGPT and any other MCP client connect with a browser sign-in. A connected agent reads a failure’s error, steps, trace and history, so it fixes the cause instead of guessing. It can also quarantine flaky tests, file Jira issues, and record before-and-after evidence of a fix for the pull request.

### How long does it take to roll out Currents?

Your first run shows up in Currents a few minutes after you install the reporter. From there, roll it out one CI pipeline at a time: add the record key to each job, invite your team or turn on SSO, and connect GitHub and Slack so results reach your pull requests and channels. Crunchbase migrated in hours, and Runa connected several Playwright projects in about a day.

### We use Cypress today. Can we move to Playwright with Currents?

Yes, and you don’t have to move all at once. Currents reports Cypress and Playwright tests side by side, so you can migrate one suite at a time. Most Currents [case studies](https://currents.dev/case-studies) are teams that made this move. See [migrating from Cypress](https://currents.dev/cypress-to-playwright).

### Is our test data secure?

Currents is SOC 2 Type II audited. It supports single sign-on over SAML 2.0 with Okta, Microsoft Entra ID and JumpCloud, user provisioning over SCIM, and roles for every member. [Data retention periods](https://docs.currents.dev/resources/data-privacy/data-retention) are published in the docs, and the [Security & Trust page](https://currents.dev/security) has the rest.

## More

- [Pricing](https://currents.dev/pricing)
- [About](https://currents.dev/about)
- [Blog](https://currents.dev/blog)
- [Case studies](https://currents.dev/case-studies)
- [Changelog](https://currents.dev/changelog)
- [Playwright dashboard](https://currents.dev/playwright)
- [Cypress to Playwright](https://currents.dev/cypress-to-playwright)
- [GitHub](https://github.com/currents-dev)
- [X](https://x.com/currents_dev)
- [LinkedIn](https://www.linkedin.com/company/currents-dev)