pxtree

pxtree

Loads a webpage in headless Chromium and prints what actually rendered as a compact text tree, so an AI coding agent can check a layout without a screenshot.

A dev server page with a row of three cards, a row of three plans, a list of steps, a row of stats, an article, a form, a column of prices and a row of tiles:

$ npx -y pxtree@latest localhost:51731280x800 light dpr 1 ltr scroll 0/382 page 1280x1182 painted to 1134since last run: no changessummary: 7 findings  a.button tops 64..136 across siblings: ul.cards  div.stat tops 0..6 across siblings: div.stats  input.field starts 0..3 across siblings: form.fields  12 wider than li.plan: li.plan.featured  72 taller than li.card: li.card 3 of 3  20 shorter than div.tile: div.tile.short  gaps 16 16 24 16 between li.step: ul.stepsbody 1280x1182 [pad 24][gaps 24 24 24 24 16 16 …, free 24 at end]  ul.cards 1232x208 [gaps across 16, free 600 at end][!! a.button tops 64..136 across siblings]    li.card 200x136 [pad 16][gaps 8][renders background]      h3 "Starter" 168x24 [text 19/24]      p "Short text." 168x24 @0,32 [text 16/24]      a.button "Choose" 168x40 @0,64 [pad 8 16][text 16/24][renders background]    li.card 200x136 @216,0 [pad 16][gaps 8][renders background]      h3 "Team" 168x24 [text 19/24]      p "Short text." 168x24 @0,32 [text 16/24]      a.button "Choose" 168x40 @0,64 [pad 8 16][text 16/24][renders background]    li.card 200x208 @432,0 [pad 16][gaps 8][renders background][!! 72 taller than li.card]      h3 "Business" 168x24 [text 19/24]      p "A much longer…" 168x96 @0,32 [text 16/24, 4 lines, 444 on one line]      a.button "Choose" 168x40 @0,136 [pad 8 16][text 16/24][renders background]  ul.plans 1232x56 @0,232 [gaps across 16, free 588 at end]    li.plan "Monthly" 200x56 [pad 16][text 16/24][renders background]    li.plan "Yearly" 200x56 @216,0 [pad 16][text 16/24][renders background]    li.plan.featured "Lifetime" 212x56 @432,0 [pad 16][text 16/24][renders background][!! 12 wider than li.plan]  ul.steps 1232x192 @0,312 [gaps 16 16 24 16][!! gaps 16 16 24 16 between li.step]    li.step "One" 1232x24 [text 16/24]    li.step "Two" 1232x24 @0,40 [text 16/24]    li.step.active "Three" 1232x24 @0,80 [text 16/24]    li.step "Four" 1232x24 @0,128 [text 16/24]    li.step "Five" 1232x24 @0,168 [text 16/24]  div.stats 1232x62 @0,528 [gaps across 16, free 456 at end][!! div.stat tops 0..6 across siblings]    div.stat "Revenue" 182x56 [pad 16][text 16/24][renders background]    div.stat "Orders" 182x56 @198,0 [pad 16][text 16/24][renders background]    div.stat.alert "Refunds" 182x56 @396,6 [pad 16][text 16/24][renders background]    div.stat "Customers" 182x56 @594,0 [pad 16][text 16/24][renders background]  article 1232x132 @0,614 [gaps 12]    h2 "First heading" 1232x24 [text 24/24]    p "First paragraph." 1232x24 @0,36 [text 16/24]    h2 "Second heading" 1232x24 @0,72 [text 24/24]    p "Second paragraph." 1232x24 @0,108 [text 16/24]  p.with-link "Read the first." 1232x24 @0,762 [text 16/24]    a "guide" 39x17 @69,3 [text 16/24]  form.fields 1232x108 @0,802 [gaps 8][!! input.field starts 0..3 across siblings]    input.field "Name" 208x21 [pad 1 2][text 13/15][renders background, border]    input.field "Email" 208x21 @0,29 [pad 1 2][text 13/15][renders background, border]    input.field.nudged "Company" 208x21 @3,58 [pad 1 2][text 13/15][renders background, border]    input.field "Phone" 208x21 @0,87 [pad 1 2][text 13/15][renders background, border]  div.prices 300x72 @0,934    span.price "$9" 18x24 @282,0 [text 16/24]    …×2 similar span.price 36x24..49x24  div.tiles 1232x80 @0,1030 [gaps across 16, free 900 at end]    div.tile 100x80 [renders background]    div.tile.short 100x60 @116,0 [renders background][!! 20 shorter than div.tile]    div.tile 100x80 @232,0 [renders background]

The first line is the facts line: viewport, color scheme, device pixel ratio, direction, scroll position out of the maximum, document size, and painted to, the lowest painted pixel. since last run compares with the previous run of the same URL and settings, or of the same --diff-key name. That history lives in ~/.cache/pxtree, never in your project, and --no-diff skips it.

The ul.cards line says the "Choose" buttons in the three cards have tops from 64 px to 136 px below the top of their card. The li.plan.featured "Lifetime" line says that plan is 212x56 at x 432 in its row, 12 px wider than the width its siblings share. Findings are measurements with a threshold, never verdicts. The agent decides from the code whether they are intended.

Why not a screenshot

• It gives numbers in CSS px, not pixels the model has to estimate.
• The agent reads it in one pass, as text, with the findings listed on top.
• One call covers several viewports and color schemes.

Install

pxtree drives Playwright's Chromium. Install it once, about 150 MB:

npx -y [email protected] install chromium

pxtree itself downloads nothing and never edits your project. --channel chrome uses an installed Chrome instead. Needs Node 20 or newer.

MCP server

OpenCode, opencode.json:

{  "$schema": "https://opencode.ai/config.json",  "mcp": {    "pxtree": { "type": "local", "command": ["npx", "-y", "pxtree@latest", "mcp"], "enabled": true }  }}

{  "mcpServers": {    "pxtree": { "command": "npx", "args": ["-y", "pxtree@latest", "mcp"] }  }}

{  "servers": {    "pxtree": { "type": "stdio", "command": "npx", "args": ["-y", "pxtree@latest", "mcp"] }  }}

{  "amp.mcpServers": {    "pxtree": { "command": "npx", "args": ["-y", "pxtree@latest", "mcp"] }  }}

{  "context_servers": {    "pxtree": { "command": "npx", "args": ["-y", "pxtree@latest", "mcp"], "env": {} }  }}

MCP Registry name: io.github.alabsi91/pxtree. The server has two tools: measure, and read_me_first, which returns what the skill holds: how to use pxtree, how to report, and the reading guide. Its instructions tell the agent to call read_me_first once per session, so no skill install is needed.

measure takes target and the CLI flags as inputs: viewports, schemes, scroll, element, children, colors, wait, script, screenshot (true saves a PNG per run under the temp directory), timeout, maxChars (default 80000), diff, diffKey, report (tree, findings, summary, changes or none) and aria (true adds the aria tree). One call can return the report, the aria tree and a screenshot together. Inputs are bounded: viewport sides 1 to 10000, at most 10 viewports, timeout at most 120000. A file target must sit under the server's working directory. That rule prevents accidents and is not a security boundary: script is trusted code that runs as Node in the server process.

Skill

For agents that read Agent Skills, Codex included through .agents/skills:

npx -y skills add alabsi91/pxtree

Or link it by hand. ~/.agents/skills is read by Codex, Cursor, Copilot, Gemini CLI, OpenCode, Amp and Devin. Claude Code reads ~/.claude/skills.

git clone https://github.com/alabsi91/pxtreemkdir -p ~/.agents/skills ~/.claude/skillsln -s "$PWD/pxtree/skills/pxtree" ~/.agents/skills/pxtreeln -s "$PWD/pxtree/skills/pxtree" ~/.claude/skills/pxtree

Use <project>/.agents/skills or <project>/.claude/skills for one project only.

Shell-only agents

Tell the agent to run this. It prints the flags, the output grammar, every tag and finding, and the limits.

npx -y pxtree@latest guide

CLI

npx -y pxtree@latest https://example.com --viewport 390x844,1280x800 --scheme light,darknpx -y pxtree@latest localhost:5173 --scroll '#pricing' --element '.card'npx -y pxtree@latest ./dist/index.htmlnpx -y pxtree@latest localhost:5173 --script "await page.click('text=Menu')" --wait '.menu'npx -y pxtree@latest localhost:5173 --element '#pricing' --screenshot pricing.pngnpx -y pxtree@latest localhost:5173 --viewport 390,1280npx -y pxtree@latest localhost:5173 --report summary --arianpx -y pxtree@latest localhost:5173 --report findingsnpx -y pxtree@latest localhost:5173 --report changesnpx -y pxtree@latest localhost:5173 --report none --screenshot shot.png

The screenshot is clipped to the --element match only when the selector matches exactly one element. Otherwise it is the whole viewport, and the facts line says how many matched.

--report picks how much of the measurement prints: tree (the default, everything), findings (the tree cut down to the lines with a finding and the names of their ancestors), summary (no tree), changes (the facts line and since last run, to check a fix) or none (the facts line only). A width alone in --viewport gets a matching height: 390x844, 768x1024, 820x1180, 1024x768, 1280x800, 1440x900, 1920x1080, and 800 for any other width. --aria adds Playwright's aria snapshot of the page, or of each --element match, after the report: roles, names, states and reading order, taken in the same state as the measurement. So one call can carry the findings, the accessibility tree and a screenshot.

npx -y pxtree@latest --help lists the flags. npx -y pxtree@latest guide explains how to read the output.

Limits

• Coverage is checked only inside the viewport.
• Declarative closed shadow roots are not walked.
• iframes are not walked. Measure the frame URL on its own.
• Pseudo-element ink is placed on the element's box.
• Clip paths count as their border box. Border radius is ignored.
• Inside a rotated or scaled box there are no overlap, centering or width findings. Its axis-aligned boxes do not say where things are.
• Contrast is not measured over images or gradients, where the [text] tag says on image, nor for a transparent text fill such as gradient text, where it says fill transparent.
• small target skips links inside a line of text and targets with enough spacing, as WCAG 2.5.8 allows.
• pxtree does not guess intent. A full-bleed section, an avatar stack, an open popover or a collapsed panel prints its numbers like anything else.
• Desktop emulation only: (hover: hover) and (pointer: fine) match. No touch.
• Vertical writing modes print physical positions.
• Loading waits at most 2 s for load and 1.5 s for a quiet network, which files skip. Slower pages need --wait.

API

import { createSession, format } from 'pxtree';
const session = await createSession();const result = await session.measure('localhost:5173', { viewports: [{ width: 390, height: 844 }] });console.log(format(result));await session.close();

License

MIT