Install into your coding agent. No account needed.
Don't see your agent? See the quick start.
Trusted by the fastest-growing companies
Teams reach reliable end-to-end coverage 10× faster because coverage grows as a byproduct of shipping, and self-healing keeps it green without the babysitting.
Browser MCP for coding agents. Auto-generate regression tests. Built-in skills for verification and reviews.
Learn morePlain YAML in your repo, describing what the user does. Review it in a PR, hand-tune it in the local debugger.
Learn moreButton moved? Class renamed? Tests auto-heal on the fly. AI Fixer repairs the rest.
Learn moreOne command locally, or Shiplight-hosted runners in CI. Results land on the pull request.
Learn moreHow it works
01Author
Your agent writes plain-English YAML tests by walking the app itself, or from a session recording, a ticket, a spec, or the test plan in a spreadsheet. Every step describes user intent, not selectors, so you review tests the way you review a spec.
Explore the YAML format> /shiplight create-yaml-tests
02Run
The same file runs locally, on Shiplight-hosted CI runners, or on your own infrastructure. Settled steps replay deterministically from cache at full Playwright speed; only new ground needs a model.
Where tests run03Heal
When a cached locator goes stale, Shiplight re-reads the page, resolves the original intent, and updates the cache mid-run. Nobody spends a morning on a renamed button, and genuine regressions still stop the line.
How self-healing works04Fix
Results land on the pull request with logs, screenshots, and traces. Ask your agent to /shiplight fix and it reproduces the failure, finds the cause, and updates the test. If the app is what broke, it reports the bug instead.
How fix works05Verify & review
The same browser skills work outside the suite. Your agent verifies a change the moment it is made and runs structured reviews for accessibility, security, and performance, with the evidence captured before code review.
See the agent workflowShiplight learns your app as it works: each run distills what it saw into grounded knowledge, so the system performs better tomorrow than it did today.
As it runs, Shiplight distills real app behavior into knowledge it can trust: your flows, your terminology, what settled and what changed.
Settled steps replay from cache with no model call. Only new or changed ground needs AI, so a mature suite runs quicker and costs less than its first run did.
Every heal, fix, and recording feeds the knowledge base. Accuracy goes up as the app evolves, instead of decaying the way hand-written suites do.
Teams reach reliable end-to-end coverage about 10x faster, with near-zero ongoing maintenance.
I used to spend 60% of my time authoring and maintaining playwright tests for our entire web application. I spent 0% of the time doing that in the past month. I'm able to spend more time on other impactful/more technical work. Awesome work!
Jeffery King
Head of QA
Within just a few days, we achieved reliable end to end coverage across our most critical flows, even with complex integrations and data driven logic. QA no longer slows the team down as we ship fast.
Binil Thomas
Head of Engineering
We automated over 80% of our core regression flows within the first few weeks. Most manual checks are gone, ongoing maintenance is minimal, and shipping changes feels significantly safer now.
Ethan Zheng
Co-founder & CTO
Test coverage across our critical AI driven flows expanded within the first month, catching multiple regressions before they reached staging. We now ship AI driven updates much more frequently without slowing engineers down.
Shuo Yang
Co-founder & CTO
Install Shiplight into your coding agent with one line. Local browser automation and test authoring need no Shiplight account and no API token, so you can verify a change you are already working on before signing up for anything. An account is what adds hosted CI runners, result history, and dashboards later.
Shiplight installs into your coding agent as an MCP server plus a set of Skills. That gives the agent eyes and hands in a real browser: it verifies UI changes as you build, then turns those verifications into end-to-end tests written as readable YAML and committed to your repository. Tests describe user intent rather than selectors, so they survive UI and DOM changes.
Your coding agent does. It walks the application, writes the YAML, and when something breaks it reproduces the failure, diagnoses the cause, and updates the test. If the application is what broke, it reports the bug instead of rewriting the test to pass. You review the result in a pull request like any other change, and you can hand-tune complex flows in the local debugger.
Shiplight is built on Playwright. Every YAML test transpiles to a standard Playwright test at run time, which is why it supports every Playwright browser and capability, and why cached steps replay at full Playwright speed. It runs alongside a Playwright suite you already have, and there is no lock-in: you can eject at any time and the transpiled tests are standard Playwright files that keep running without Shiplight. Selenium is not involved anywhere.
No. Tests are plain-English YAML written and maintained by your coding agent, so there are no selectors to write and no Playwright code to keep working. Playwright is the execution engine underneath: you get its speed and reliability without its learning curve. When the UI changes, Shiplight re-reads the page and re-resolves the step's intent instead of you rewriting locators.
Mechanical drift heals itself: if a cached locator goes stale, Shiplight re-reads the page, resolves the original intent, and updates the cache mid-run. For real failures, results land on the pull request with logs, screenshots, and traces, and your agent can reproduce the failure, find the cause, and update the test. If the application is what broke, it reports the bug instead of forcing the test to pass.
Locally with a single command, on Shiplight-hosted CI runners, or in your own CI. The same YAML files run in all three, so moving between them changes a workflow file rather than your suite. You can bring your own LLM key and run on your own infrastructure on any plan, including the free one.