Screenshot Local - Capture Project Screenshots with shot-scraper

Overview

Capture screenshots of local development projects using shot-scraper installed globally via pipx. Turn localhost URLs and local HTML files into PNGs, JPEGs, and PDFs for documentation, READMEs, and design reviews.

Core principle: Screenshots should be reproducible from a single command or YAML config, not manual screen captures. Version-control your screenshot configs alongside your code.

When to Use

Use this skill when:

• Capturing screenshots of a local dev server for docs or README
• Generating screenshots of multiple pages/states in batch
• Documenting UI changes or new features visually
• Creating before/after comparisons during refactoring
• Automating screenshot generation in CI/CD

Avoid when:

• Recording terminal/CLI output (use record-tui with VHS instead)
• Capturing non-web content (use native screenshot tools)
• Scraping third-party websites (shot-scraper can do it, but this skill focuses on local projects)

Prerequisites

Install shot-scraper globally via pipx:

# Install pipx if needed
brew install pipx  # macOS
# or: apt install pipx / pip install pipx

# Install shot-scraper globally
pipx install shot-scraper

# Install the browser engine (Chromium by default)
shot-scraper install

Verify: shot-scraper --version

Alternative browsers:

shot-scraper install -b firefox
shot-scraper install -b webkit

Phase 1: Plan the Screenshots

Before capturing, answer these questions:

1. What are you capturing? — localhost URL, local HTML file, or built static site
2. Which pages/states? — Single page, multiple routes, or specific UI states
3. What dimensions? — Match your target display context
4. Any interaction needed? — Click buttons, fill forms, dismiss modals before capture

Recommended dimensions by use case:

Phase 2: Capture Screenshots

Single Screenshot

Capture a localhost page:

shot-scraper http://localhost:3000 -o homepage.png

Capture a local HTML file:

shot-scraper index.html -o preview.png

With custom dimensions:

shot-scraper http://localhost:3000 -w 1200 -h 800 -o homepage.png

Retina (2x resolution):

shot-scraper http://localhost:3000 --retina -o homepage-retina.png

Capture a specific element:

shot-scraper http://localhost:3000 -s ".hero-section" -o hero.png
shot-scraper http://localhost:3000 -s "#main-nav" --padding 10 -o nav.png

Wait for dynamic content:

# Wait fixed time (ms)
shot-scraper http://localhost:3000 --wait 2000 -o page.png

# Wait for JS condition
shot-scraper http://localhost:3000 --wait-for "document.querySelector('.loaded')" -o page.png

Execute JS before capture (dismiss modals, set state):

shot-scraper http://localhost:3000 \
  -j "document.querySelector('.cookie-banner')?.remove()" \
  -o clean-homepage.png

Essential Options

For the complete flag reference, see Command Reference.

Batch Screenshots with YAML

Create a shots.yml for capturing multiple pages:

# shots.yml
- url: http://localhost:3000
  output: screenshots/homepage.png
  width: 1280
  height: 800

- url: http://localhost:3000/about
  output: screenshots/about.png
  width: 1280

- url: http://localhost:3000/dashboard
  output: screenshots/dashboard.png
  width: 1280
  height: 900
  wait: 2000
  javascript: |
    document.querySelector('.loading-spinner')?.remove()

- url: http://localhost:3000
  output: screenshots/mobile-home.png
  width: 375
  height: 812

Run the batch:

shot-scraper multi shots.yml

Advanced YAML Keys

- url: http://localhost:3000/form
  output: screenshots/form-filled.png
  width: 1200
  javascript: |
    document.querySelector('#name').value = 'Jane Doe';
    document.querySelector('#email').value = 'jane@example.com';
  wait: 500
  selector: ".form-container"
  padding: 20
  retina: true

Supported YAML keys: url, output, width, height, quality, wait, wait_for, selector, selectors, selector_all, padding, javascript, js_selector, retina, omit_background

Phase 3: Smart Config Generation

When asked to generate a screenshot config for a project, follow this process:

Step 1: Analyze the Project

Read the project to understand:

• Routes/pages — Check router config, page files, or navigation
• Key UI states — Loading, empty, populated, error states
• Important components — Hero sections, dashboards, forms
• Port — What port does the dev server run on

Step 2: Generate shots.yml

Build a YAML config covering the project's key screens:

# Validate by running
shot-scraper multi shots.yml

Step 3: Iterate

Review screenshots, adjust timing and selectors, re-run.

Phase 4: Optimize Output

File Size Reduction

Use JPEG for photos/complex UIs:

shot-scraper http://localhost:3000 --quality 85 -o page.jpg

Use PNG for UI with text/sharp edges (default).

Transparent backgrounds (for overlaying on docs):

shot-scraper http://localhost:3000 -s ".component" --omit-background -o widget.png

Consistency Tips

• Always set explicit width — default 1280 is fine but be intentional
• Omit height for full-page captures, set it for fixed viewport
• Use --retina for README images viewed on high-DPI screens
• Add --wait for SPAs that hydrate client-side
• Use javascript to dismiss cookie banners, tooltips, or modals
• Use selector + padding to capture specific components cleanly

Phase 5: CI/CD Integration

GitHub Actions

name: Update Screenshots
on:
  push:
    branches: [main]
    paths: ["src/**", "shots.yml"]

jobs:
  screenshots:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Setup Python
        uses: actions/setup-python@v5
        with:
          python-version: "3.12"

      - name: Install shot-scraper
        run: |
          pip install shot-scraper
          shot-scraper install

      - name: Capture screenshots
        run: |
          npm ci && npm run dev &
          sleep 5
          shot-scraper multi shots.yml

      - name: Commit updated screenshots
        uses: stefanzweifel/git-auto-commit-action@v5
        with:
          commit_message: "docs: update screenshots"
          file_pattern: "screenshots/*.png"

Tip: Use the server key in YAML to auto-start a dev server — see Templates.

Examples

Good: Well-Structured Screenshot Config

# shots.yml — explicit dimensions, proper waits, organized output
- url: http://localhost:3000
  output: screenshots/homepage.png
  width: 1280
  height: 800

- url: http://localhost:3000/dashboard
  output: screenshots/dashboard.png
  width: 1280
  height: 900
  wait: 2000
  javascript: |
    document.querySelector('.toast-notification')?.remove()

- url: http://localhost:3000/settings
  output: screenshots/settings.png
  width: 1280
  selector: ".settings-panel"
  padding: 20

- url: http://localhost:3000
  output: screenshots/mobile-home.png
  width: 375
  height: 812

Bad: Common Mistakes

# No output specified — shot-scraper auto-names from URL, messy results
- url: http://localhost:3000

# No width — relies on default, inconsistent across machines
- url: http://localhost:3000/about
  output: about.png

# No wait for SPA — captures loading spinner instead of content
- url: http://localhost:3000/dashboard
  output: dashboard.png

# Screenshot directory not organized
- url: http://localhost:3000/settings
  output: settings.png

Good: Capturing Specific Components

# Hero section with padding for breathing room
shot-scraper http://localhost:3000 -s ".hero" -p 20 -o docs/hero.png

# Navigation in retina for crisp text
shot-scraper http://localhost:3000 -s "nav" --retina -o docs/nav.png

# Form with pre-filled data via JS
shot-scraper http://localhost:3000/contact \
  -j "document.querySelector('#name').value = 'Jane Doe'" \
  -s ".contact-form" -p 10 -o docs/form.png

Bad: Component Capture Anti-Patterns

# No selector — captures entire page when you only need one section
shot-scraper http://localhost:3000 -o hero.png

# No padding — element cropped tight, looks cramped in docs
shot-scraper http://localhost:3000 -s ".hero" -o hero.png

# No wait — dynamic component hasn't rendered yet
shot-scraper http://localhost:3000 -s ".chart" -o chart.png

Troubleshooting

Problem: Screenshot shows blank page or loading spinner

Cause: SPA hasn't hydrated or async data hasn't loaded.

Solution:

# Fixed wait
shot-scraper http://localhost:3000 --wait 3000 -o page.png

# Wait for specific element
shot-scraper http://localhost:3000 --wait-for "document.querySelector('.content')" -o page.png

Problem: "Connection refused" on localhost

Cause: Dev server isn't running or is on a different port.

Solution:

1. Start your dev server first: npm run dev &
2. Verify the port: curl -I http://localhost:3000
3. Or use the server key in YAML to auto-start it

Problem: Screenshot dimensions don't match expectations

Cause: Height omitted (captures full page) or retina not accounted for.

Solution:

• Set explicit -h for fixed viewport
• Omit -h intentionally for full-page capture
• --retina doubles pixel dimensions (1280w becomes 2560px image)

Problem: shot-scraper command not found

Cause: Not installed or pipx PATH not configured.

Solution:

pipx install shot-scraper
pipx ensurepath      # Add pipx bin to PATH
shot-scraper install # Install browser engine

Problem: Elements missing from screenshot

Cause: CSS selector wrong, element behind a modal, or lazy-loaded content.

Solution:

# Use --interactive to debug visually
shot-scraper http://localhost:3000 -i

# Use --devtools for inspector
shot-scraper http://localhost:3000 --devtools

Integration

This skill pairs with:

• record-tui — For terminal app recordings (use VHS for CLIs, shot-scraper for web UIs)
• track-session — Track screenshot iteration progress
• git-worktree — Compare screenshots across branches

Useful alongside:

• shot-scraper — The screenshot tool
• shot-scraper-template — GitHub Actions template
• pipx — Global Python app installer

Quick Reference

# Single screenshot
shot-scraper http://localhost:3000 -o page.png

# With dimensions
shot-scraper http://localhost:3000 -w 1280 -h 800 -o page.png

# Specific element
shot-scraper http://localhost:3000 -s ".hero" -p 10 -o hero.png

# Retina
shot-scraper http://localhost:3000 --retina -o page@2x.png

# Local HTML file
shot-scraper index.html -o preview.png

# Batch from YAML
shot-scraper multi shots.yml

# Interactive debugging
shot-scraper http://localhost:3000 -i

# PDF export
shot-scraper pdf http://localhost:3000 -o page.pdf

Deep Reference

For detailed guides, load these files when needed:

• Command Reference — All shot-scraper commands, flags, and subcommands
• Templates — Copy-paste YAML configs for common project types

Only load these when specifically needed to save context.

References

• shot-scraper — Official repository
• shot-scraper documentation — Full docs
• shot-scraper-template — GitHub Actions template
• pipx — Global Python app installer
• Simon Willison's blog — Author's posts on shot-scraper