Agent-Skills.md

Agent Skills: Vizzly Visual Testing Knowledge Base

Core knowledge about Vizzly visual testing - file structure, CLI commands, and how it works. Use when working with Vizzly, running visual tests, or helping with screenshot-related tasks.

UncategorizedID: vizzly-testing/cli/vizzly-knowledge

Repository

vizzly-testing/cli
vizzly-testingLicense: MIT
71

Install this agent skill to your local

pnpm dlx add-skill https://github.com/vizzly-testing/cli/tree/HEAD/vizzly-plugin/plugins/vizzly/skills/vizzly-knowledge

Skill Files

Browse the full folder contents for vizzly-knowledge.

Download Skill

Loading file tree…

vizzly-plugin/plugins/vizzly/skills/vizzly-knowledge/SKILL.md

Skill Metadata

Name
vizzly-knowledge
Description
Core knowledge about Vizzly visual testing - file structure, CLI commands, and how it works. Use when working with Vizzly, running visual tests, or helping with screenshot-related tasks.

Vizzly Visual Testing Knowledge Base

You have deep knowledge of Vizzly, a visual regression testing platform. This context helps you work effectively with Vizzly projects.

What is Vizzly?

Vizzly captures screenshots from real functional tests (not isolated component renders) and compares them against baselines. It works in two modes:

TDD Mode (Local Development):

  • Run vizzly tdd start to start the local server
  • Screenshots are compared locally using honeydiff (fast Rust-based diffing)
  • Interactive dashboard at http://localhost:47392
  • Baselines stored in .vizzly/baselines/
  • No API token required

Cloud Mode (CI/CD):

  • Run vizzly run "npm test" --wait
  • Screenshots uploaded to Vizzly cloud
  • Team reviews changes via web dashboard
  • Requires VIZZLY_TOKEN

The .vizzly Directory

When Vizzly runs, it creates a .vizzly/ directory with this structure:

.vizzly/
├── baselines/          # Baseline screenshots (PNG files)
│   ├── homepage.png
│   └── login-form.png
├── current/            # Current test screenshots
│   ├── homepage.png
│   └── login-form.png
├── diffs/              # Visual diff images (red highlights)
│   └── homepage.png
├── report-data.json    # Comparison results (JSON)
├── server.json         # TDD server info (port, etc.)
└── baseline-metadata.json  # Source of baselines (optional)

Understanding report-data.json

This is the primary source of truth for test results. Read it to understand test status:

{
  "comparisons": [
    {
      "name": "homepage",
      "status": "failed",        // "passed", "failed", or "new"
      "diffPercentage": 2.3,     // Percentage of pixels different
      "threshold": 0.1,          // Allowed diff percentage
      "current": "/images/current/homepage.png",
      "baseline": "/images/baselines/homepage.png",
      "diff": "/images/diffs/homepage.png"
    }
  ]
}

Status meanings:

  • passed - Screenshot matches baseline within threshold
  • failed - Screenshot differs more than threshold allows
  • new - No baseline exists yet (first time this screenshot was captured)

CLI Commands

TDD Mode (Local Development):

vizzly tdd start              # Start TDD server (background)
vizzly tdd run "npm test"     # Run tests with ephemeral server
vizzly tdd status             # Show current test status
vizzly tdd stop               # Stop the TDD server
vizzly baselines              # List and query local baselines

Cloud Mode (CI/CD):

vizzly run "npm test"         # Run tests and upload to cloud
vizzly run "npm test" --wait  # Wait for cloud processing
vizzly status <build-id>      # Check build status
vizzly builds                 # List and query builds
vizzly finalize <parallel-id> # Finalize parallel builds
vizzly comparisons            # Query and search comparisons

Review Commands (approve/reject visual changes):

vizzly comparisons -b <build-id>                     # List comparisons for a build
vizzly comparisons --name "Button*" --status changed # Search by name/status
vizzly approve <comparison-id>                       # Approve a comparison
vizzly approve <comparison-id> -m "LGTM"             # Approve with comment
vizzly reject <comparison-id> -r "reason"            # Reject (reason required)
vizzly comment <build-id> "message"                  # Add comment to a build

Project Setup:

vizzly init                   # Create vizzly.config.js
vizzly config                 # Display current configuration
vizzly doctor                 # Run diagnostics to check environment

Account & Authentication:

vizzly login                  # Authenticate with Vizzly cloud
vizzly logout                 # Clear stored authentication tokens
vizzly whoami                 # Show current auth status
vizzly orgs                   # List organizations you have access to
vizzly projects               # List projects you have access to

Project Configuration:

vizzly project:select         # Configure project for current directory
vizzly project:list           # Show all configured projects
vizzly project:token          # Show project token for current directory
vizzly project:remove         # Remove project configuration

Advanced:

vizzly api <method> <endpoint>  # Make raw API requests
vizzly upload                   # Upload screenshots directly
vizzly preview                  # Upload static files as preview

Taking Screenshots in Tests

The client SDK provides vizzlyScreenshot():

import { vizzlyScreenshot } from '@vizzly-testing/cli/client';

// In your test
await vizzlyScreenshot('homepage', await page.screenshot(), {
  browser: 'chrome',
  viewport: '1920x1080'
});

Screenshot Identity: Screenshots are matched by signature: name|viewport_width|browser. The same logical screenshot across runs can be compared even with different parameters.

Accepting Baselines

When a screenshot changes intentionally, you need to accept it as the new baseline:

Via TDD Dashboard (Local):

  1. Open http://localhost:47392
  2. Review the visual diff
  3. Click "Accept" on screenshots you want to update

Via CLI (Cloud builds):

vizzly comparisons -b <build-id>          # List comparisons for a build
vizzly approve <comparison-id>            # Approve a comparison
vizzly reject <comparison-id> -r "reason" # Reject (reason required)

Via File Operations (Local TDD): Copy the current screenshot to baselines:

cp .vizzly/current/homepage.png .vizzly/baselines/homepage.png

Configuration

Vizzly uses vizzly.config.js for configuration:

import { defineConfig } from '@vizzly-testing/cli/config';

export default defineConfig({
  threshold: 0.1,           // Default diff threshold (0.1 = 0.1%)
  port: 47392,              // TDD server port
  basePath: '.vizzly',      // Where to store screenshots
});

Environment Variables

  • VIZZLY_TOKEN - API authentication token
  • VIZZLY_API_URL - API base URL (default: https://app.vizzly.dev)
  • VIZZLY_ENABLED - Enable/disable SDK (default: auto-detect)
  • VIZZLY_LOG_LEVEL - Logging level (debug|info|warn|error)

Common Workflows

Starting visual testing on a project:

  1. Install: npm install --save-dev @vizzly-testing/cli
  2. Init: npx vizzly init
  3. Add screenshots to tests
  4. Run: vizzly tdd run "npm test"
  5. Review and accept baselines

Debugging a failing screenshot:

  1. Read .vizzly/report-data.json to find the failing comparison
  2. View the baseline and current images
  3. Compare visually to identify what changed
  4. Decide: accept as new baseline OR fix the visual regression