Browser Use
Lets AI do web tasks for you, such as filling forms, searching and extracting data, in the maker’s cloud browsers or on your computer.
Description from the maker (original English)
Open HTML reports in the browser and edit them live while the person watches, or by hand.
These three are never merged into one score: opinions, measurements and popularity are different things.
We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.
Pick the AI app you use, then copy the settings.
You need
Which app do you use?
Pick your app above. We remember it, so next time the button says “Install in …”.
Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json
{
"mcpServers": {
"hotedit": {
"command": "npx",
"args": [
"-y",
"hotedit",
"mcp"
]
}
}
}Run this in your terminal
claude mcp add hotedit -- npx -y hotedit mcpPaste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"hotedit": {
"command": "npx",
"args": [
"-y",
"hotedit",
"mcp"
]
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
Paste into .vscode/mcp.json in your project
{
"servers": {
"hotedit": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"hotedit",
"mcp"
]
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.
Your choice stays on this device in a cookie; no account needed.
These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.
Runs on your computer; the tool list appears in your AI app after you install it.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 7, 2026
The registry name was verified with the GitHub account Thewhey-Brian, which owns the repo
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
This one runs on your computer; there is no internet endpoint
Runs on your computer and uses local settings instead
Runs on your computer; we can only read tool lists from internet endpoints
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).
The GitHub repo last changed –, and the latest version (0.5.0) reached the registry 15 days ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.
In the same category: Browser Use, Chrome DevTools MCP, MCP Server Browser
Servers in the same category from other makers, sorted by stars.
Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
Lets AI do web tasks for you, such as filling forms, searching and extracting data, in the maker’s cloud browsers or on your computer.
Lets AI use Chrome’s developer tools to inspect pages, find what makes them slow and spot errors. Made by the Chrome DevTools team.
MCP server for browser use access
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
Edit any HTML file in your browser — text, styles, layout and charts — and save the changes straight back into the file. Built for the HTML reports that AI assistants write.
▶ Watch the narrated 95-second tour (MP4, sound on) · or try it yourself: npx hotedit demo
AI assistants produce a lot of HTML: reports, dashboards, summaries. They look great, but fixing a typo, renaming a chart title or moving a section means digging through thousands of lines of generated markup, or asking the AI again. hotedit opens the page in your browser so you can click what you see and change it. Only the bytes you changed are rewritten, so the file stays clean, diff-friendly and yours.
It doesn't change what your AI makes. Let it write the report exactly as it normally would, at the same quality and with the same charts, design and depth. hotedit only adds the ability to edit it afterwards. Nothing is simplified or re-generated, and nothing is added to the file.
You need Node.js 18+. Then:
npx hotedit report.html
Your browser opens the page. Click ✎ Edit (or press ⌥E / Alt+E), click any text and type. Changes are saved into report.html as you go, ⌘Z / Ctrl+Z undoes them, and ✓ Done (top right, always in the same place) saves anything still open and stops editing. Hover any button to see its shortcut: they are all ⌥ / Alt + a key (⌥E edit, ⌥A ask AI, ⌥S style, ⌥I add, ⌥D duplicate, ⌥⌫ delete, ⌥↑/↓ move), so plain keys never trigger anything and typing is never affected.
Just want to see it? Open a sample report (charts, maths, diagrams) in a throwaway copy:
npx hotedit demo
Not sure which file? Run it with no file and click one from the list:
npx hotedit # lists the HTML files in this folder
To install it permanently, so you can just type hotedit:
npm install -g hotedit
hotedit report.html
| How | |
|---|---|
| Edit text | Click and type; the caret lands where you click. Click the next block to save and move on; Enter saves, Esc cancels, ⌘B / ⌘I for bold and italic. In a list, Enter at the end of an item starts a new one. |
| Add things | Select a block and click +: paragraph, heading, list or list item, box, quote, table, code, divider, image, chart, Mermaid diagram or your own HTML. Choose above, below or inside. |
| Move things | Drag ⠿ to move a block anywhere: list items into another list, a paragraph into a box. Or use ⋯ → Move up / down (Alt+↑ / Alt+↓). |
| Style | Aa: size, weight, colour, alignment, font, spacing, background, corners and border. |
| Charts | Click a Plotly, Chart.js, Vega-Lite or Mermaid chart. Pick a type from the tiles (bar, line, area, dots, pie, donut…) and it redraws instantly. Style has switches, menus and colour pickers for titles, axes, legend and series; Data is an editable table with + row; Code is the full spec. Everything previews live; Apply saves. |
| Images & links | 🖼 replaces an image, 🔗 changes a link. |
| Everything else | ⋯: duplicate, hide, delete, or Edit HTML for the selected block. The name chip (e.g. *Paragraph ▴*) selects the block around it. |
| Ask AI, live | ✨ on a block (or ✨ AI for the whole page): type what should change, and the AI edits the page while you watch, in place, without reloading. Select a block first and it changes only that block. Every change glows as it lands and Undo AI changes takes them all back. It's a chat: ask follow-ups in the box at the bottom, and it shows which AI and model answered. |
Things that are generated by the page's own scripts, such as KaTeX maths, Mermaid output or tables built in JavaScript, are never saved as rendered output. hotedit opens their *source* instead.
Click ✨, say what you want ("make this shorter", "turn the bars orange", "add a row for March"), and press Do it. The AI reads the page, makes small edits in the file and each one shows up in your browser straight away, without reloading. Connect it once with ⚙, using an AI you already have:
claude once to log in.npm install -g @openai/codex and run codex once to log in.~/.htmledit/config.json), or set ANTHROPIC_API_KEY. Uses claude-opus-5-5 unless you pick another model.gemini mcp add hotedit npx -y hotedit mcp, type gemini -y. The request is added as the last argument (or wherever you put {prompt}).Change just one thing. With a block selected, the panel says *Only changes: …* and the AI can edit that block only (for a chart, also the code that draws it). Anything else it tries is rejected. Click another block to point it there, or Whole page to lift the limit. All its changes are normal undo steps.
Give any coding agent that speaks MCP (Claude Code, Codex, Cursor, Gemini CLI…) hotedit's tools, and it can open a report in your browser and edit it while you watch:
claude mcp add hotedit -- npx -y hotedit mcp
codex mcp add hotedit -- npx -y hotedit mcp
Tools: open (opens the file in hotedit and your browser), outline, read and edit (saved and shown live). The agent finds pages you already have open in hotedit, too.
Install it once. Your agent then writes reports that are easy to edit (text in HTML, chart specs inline), and every HTML report it writes opens in hotedit in your browser automatically, ready to edit. Later edits by the agent show up on the page live.
As a Claude Code plugin (skill, auto-open hook and MCP tools in one go):
/plugin marketplace add Thewhey-Brian/hotedit
/plugin install hotedit@hotedit
Or from the command line:
npx hotedit skill install # Claude Code, all projects (~/.claude)
npx hotedit skill install --project # Claude Code, this project only (./.claude) npx hotedit skill install --codex # Codex (~/.codex)
This adds the skill and a small hook that runs after the agent writes a file. It ignores files that aren't HTML, pages that belong to an app or website (under src/, public/, node_modules/… or with template tags), and pages already open. The page is served in the background, and the server stops by itself once you've closed it for a while. Your other settings and hooks are left as they are.
--no-auto. Pause it for one session: HOTEDIT_AUTO=0.npx hotedit skill uninstall (add --codex or --project to match).npx hotedit skill print (paste it into AGENTS.md). They open a report with npx -y hotedit open report.html --background, which returns at once.The skill only shapes *how* the HTML is written: text in the markup, chart specs inline. It tells the agent never to simplify or leave anything out for the sake of editing.
report.html ──parse (keeps source positions)──► served copy with element ids + editor overlay
▲ │ └──────── small byte patch ◄── map your edit back to source ◄──┘
·, quote styles, indentation and comments elsewhere stay exactly as they were. is patched value by value, keeping its own quote and key style.~/.htmledit/backups/).127.0.0.1 and checks a per-session token.Does it work on any HTML file? Yes, any local .html file: hand-written, AI-generated, or exported from Jupyter, plotly.py, R Markdown or Quarto. It is tested on pages with KaTeX, Mermaid, inline SVG charts, JavaScript-built dashboards, base64 images and deliberately odd markup.
Will it mess up my file? It is designed not to. Every test page is opened, every editable element is saved unchanged, and the file must stay byte-identical. Each real edit must produce exactly the expected diff.
Does it need the internet? No, hotedit runs locally. Your page may load its own libraries from a CDN, as many AI reports load KaTeX or Plotly.
Which browsers? Any modern browser. Automated tests run in Chromium.
Can I edit content that a script builds? Use ⋯ → Edit HTML on the script's block, or ask the AI with ✨.
.css and .js files are not edited, and styles are written as inline style attributes. (directly, or through const spec = {...}). Specs computed at runtime or loaded from separate files open as source.git clone https://github.com/Thewhey-Brian/hotedit && cd hotedit
npm install npx playwright install chromium # browser for the end-to-end tests npm test # unit, CLI and real-browser tests (~3 min) node demo/audio.mjs # tour narration, sound effects, music (ElevenLabs key) node demo/record.js # re-record the tour video + GIF (needs ffmpeg)
See CONTRIBUTING.md and docs/ARCHITECTURE.md.
hotedit is young and contributions are very welcome. Good places to start:
Roadmap: more chart libraries (ECharts, Observable Plot, Bokeh), editing external CSS, a VS Code extension, and guides for more agents (Gemini CLI, Cursor, OpenCode).
document.designMode saves whatever the scripts rendered.MIT