Shiplight AI Services: The Verification Stack Built for AI-Native Shipping

# Shiplight AI Services: The Verification Stack Built for AI-Native Shipping

AI coding agents have changed the pace of product development. They can generate UI changes quickly, open pull requests continuously, and refactor large surfaces of an application in a single session. The bottleneck has moved upstream: it is no longer implementation speed, it is confidence.

Shiplight AI is built for that reality. It plugs verification into the development loop, captures what was verified as durable regression coverage, and keeps that coverage stable as the UI evolves. The platform is not one feature dressed up as a category. It is a connected set of services designed to solve a single problem end-to-end: **ship faster without turning QA into a maintenance tax**.

Below is a practical guide to Shiplight AI’s core services, what each includes, who it is for, and the value it delivers.

---

## Shiplight Plugin for AI coding agents

When development is agentic, verification needs to happen in the same loop where changes are made, not hours later in CI. Shiplight Plugin connects AI coding agents to a real browser using a Browser MCP server, then layers in “skills” that encode proven QA workflows.

**What it includes**
- **Browser MCP server** so an agent can see, click, type, and navigate like a real user.
- **Slash-command skills** for repeatable workflows, including:
- `/verify` for validating UI changes in a real browser
- `/create_e2e_tests` for generating reusable E2E coverage
- `/review` for automated reviews (including security and accessibility) that can also generate regression tests from findings
- `/cloud` to sync work into Shiplight Cloud for scheduled runs and CI

**Who it’s for**
- Teams actively using AI coding agents (Claude Code, Cursor, Codex, GitHub Copilot) who want verification to keep up with throughput.

**Value**
- Faster feedback while code is still fresh in context.
- Verification that produces an artifact you can keep: regression coverage that grows with every feature shipped.

---

## Intent-based YAML tests that stay readable and self-heal

Shiplight’s YAML E2E test format is designed around intent: tests describe what the user is trying to do, not brittle implementation details. Locators can be cached for speed, and when the UI changes, Shiplight can re-derive the correct target from intent instead of forcing manual selector repair.

**What it includes**
- **Plain YAML test definitions** that “read like user stories,” making tests reviewable in pull requests and understandable beyond QA.
- **Intent-driven execution** that reduces breakage from DOM refactors and UI rearrangements.
- **Compatibility with Playwright**: Shiplight runs on top of Playwright, keeping execution grounded in real browser behavior.

**Who it’s for**
- Engineering teams that want tests to live alongside code and be part of normal review workflows.
- Cross-functional teams where PMs, designers, and QA should be able to read and contribute to coverage.

**Value**
- Less translation between “what we meant” and “what the test asserts.”
- A test suite that stays stable through UI churn, without full-time babysitting.

---

## Visual Test Editor with recording and AI-assisted authoring

Not every team wants to author tests purely as files, and not every workflow is fastest in an IDE. Shiplight Cloud includes a visual editor where AI can draft flows from goals or natural language steps, and teams can refine them visually.

**What it includes**
- **Test creation from goals**: provide a goal and Shiplight can generate a complete flow draft (navigation, interactions, verifications).
- **Recording**: capture steps by interacting with the browser while the debugger is active, then save those steps into the test.
- **AI-powered assertions** that understand common verification language (content, state, navigation checks).

**Who it’s for**
- QA teams modernizing away from script-heavy frameworks.
- Product teams that want tighter collaboration on what “done” means, expressed directly as executable checks.

**Value**
- Rapid coverage creation without requiring everyone to become a test-framework specialist.
- A smoother path from “we validated it once” to “we will never regress this again.”

---

## Shiplight Cloud for runners, schedules, and evidence-rich results

Once tests exist, the next question is operational: how do you run them continuously and get signal you can act on?

Shiplight Cloud stores test cases, triggers runs across environments, and provides the artifacts teams need to debug quickly: step-level status, screenshots, runner logs, and more.

**What it includes**
- **Cloud execution and results** with detailed run data and downloadable artifacts.
- **Schedules** (recurring runs via cron expressions) that can include suites and individual test cases, with reporting on results and performance metrics.
- **Deep debugging views** including screenshots per step, video playback, Playwright trace files, network activity, DOM snapshots, and console logs.
- **AI Test Summary** that analyzes failed runs to provide root-cause identification, human-readable explanations, and visual context analysis from screenshots.

**Who it’s for**
- Teams that need shared visibility across environments (preview, staging, production-like).
- Organizations that want E2E to become a dependable release signal, not a flaky afterthought.

**Value**
- Faster time-to-diagnosis when something breaks.
- Continuous coverage that does not depend on heroic manual triage.

---

## CI and workflow integrations that make quality enforceable

Reliability only matters if it shows up where decisions get made: pull requests, deployments, and incident workflows.

**What it includes**
- **GitHub Actions integration** using a Shiplight GitHub Action that takes an API token plus suite and environment IDs, and supports running multiple suites in parallel.
- **Webhooks** that publish test results when runs complete, with subscription conditions such as “failed only” or “pass→fail regressions,” making it easy to plug into internal tooling and notifications.
- **Hooks** (before/after) that run setup and teardown templates automatically, reducing duplication across tests and centralizing common prep and cleanup.

**Who it’s for**
- Platform and DevOps teams responsible for CI quality gates.
- Engineering orgs that want consistent release hygiene without adding process overhead.

**Value**
- Automated checks that are enforceable, visible, and integrated into the systems teams already use.

---

## Developer experience options: VS Code extension and Desktop App

Shiplight is built to reduce context switching. That includes how you debug.

**What it includes**
- **VS Code Extension** that brings an interactive visual debugger into the editor for `.test.yaml` files, so engineers can step through statements, inspect and edit action entities inline, and iterate quickly.
- **Native macOS Desktop App** (Apple Silicon) that loads the Shiplight web UI while running the browser sandbox and AI agent worker locally for fast debugging without cloud browser sessions. It supports bring-your-own AI provider keys (stored in macOS Keychain) and includes a bundled MCP server.

**Who it’s for**
- Teams that want the fastest possible inner loop for verification and debugging.
- Engineers who prefer IDE-native workflows and local, headed browser visibility.

**Value**
- Less friction in the moments that usually slow teams down: reproducing, debugging, and fixing.

---

## Enterprise readiness for mission-critical workflows

When E2E testing touches production-like data, credentials, or regulated workflows, “we’ll address security later” is not a plan.

Shiplight positions enterprise readiness explicitly, including SOC 2 Type II compliance, a 99.99% uptime SLA, and private cloud and VPC deployment options.

**Who it’s for**
- High-growth teams scaling into enterprise requirements.
- Enterprises that need clear compliance posture, auditability, and deployment flexibility.

**Value**
- Confidence that your verification layer can be treated as production-grade infrastructure.

---

## Where to start

Most teams adopt Shiplight in the order that matches how they ship:

- Start with **Shiplight Plugin** to verify changes in a real browser as code is written.
- Convert those verifications into **intent-based YAML regression tests** you can review and keep.
- Scale execution with **Shiplight Cloud**, schedules, and CI integrations.
- Tighten the loop with the **VS Code extension** or **Desktop App** when speed and debugging matter most.

If you want Shiplight AI to map cleanly to your workflow, the best next step is to pick one critical user journey, verify it in a real browser, and let that verification become your first piece of durable regression coverage.