See where your Wi-Fi is weak.
Walk around with a laptop and click where you stand on your floor plan: Wi-Fi Heatmapper measures the Wi-Fi signal there, and optionally the real speed. A few rooms later you have a Wi-Fi coverage map of your home or office: heat maps that show the dead zones, which rooms need a mesh node or an extender, and whether a slow connection is really the Wi-Fi's fault. A free, open-source (MIT) Wi-Fi site survey tool for macOS, Windows and Linux.
Website and guide: https://wifi-heatmapper.danielalder.cz
| System | Get it |
|---|---|
| macOS 13.5 (Ventura) or later | .dmg for Apple silicon or Intel from the latest release, or brew install --cask hnykda/tap/wifi-heatmapper (installs iperf3 too) |
| Windows 10 and 11, x64 | -setup.exe (just you) or .msi (everyone) from the latest release |
| Linux x64 and ARM64 | .deb or AppImage from the latest release |
| Docker (Linux hosts) | docker build -t wifi-heatmapper ., then run it with host networking |
| From source | Node.js 22.12 or newer: git clone https://github.com/hnykda/wifi-heatmapper.git, npm install, npm run dev (more) |
The apps are not signed yet, so macOS and Windows warn you the first time: Installing the app shows what to click. On an older Mac, run it from source.
- Settings: pick the bundled floor plan or upload your own. A phone photo of a pencil sketch is fine.
- What to measure: leave it at Signal only, or choose Signal and speed (see below).
- Floor plan: stand somewhere and click that spot. A coloured dot appears. Do this in every room.
- Heat maps: drag the radius until the dots merge into one map, then download the maps you need.
Getting started has the details.
Signal only needs nothing else and takes about 2 seconds a spot: it shows where coverage is weak. Signal and speed also measures download and upload speed against a second computer wired to your router (with iperf3), so you see what you actually get. See Measuring speed.
An optional MCP server lets Claude Code, Claude Desktop or another agent read and explain your surveys, set up a new one, and walk you through measuring it while you carry the laptop: it says where to go, measures once you confirm, and tells you the result. It only talks to the app on your computer and never sees your sudo password.
npm run mcp:build
claude mcp add wifi-heatmapper -- node "$PWD/dist/mcp/wifi-heatmapper-mcp.mjs"The desktop app ships it too. Setup for every client: Using it with an AI agent.
Your surveys stay on your computer, as plain JSON files. The app sends only anonymous usage counts (installs, app starts, number of measurements; never network names, addresses or results): I build it on my own, and they tell me it's used. Turn them off in Settings any time, or see exactly what is sent in Privacy and usage counts.
Wi-Fi works differently on every system and version, and I can't test them all, so your report really helps. In the app, click the ? in the bottom-right corner and choose Report a problem: it opens a GitHub issue with your system and the app's diagnostics filled in, with the sudo password, network names and addresses left out (what's in it). Tried a desktop app and it just worked? Tell us that too.
- Getting started: install and your first survey
- Installing the app: per system, first launch, data and log folders, what has been tested
- Using the app: every tab and setting
- Measuring speed: the second computer and iperf3
- How it works: where readings come from, percent vs dBm, noise, the heat map
- Running from source: Node.js, mock mode, Docker, logs
- Using it with an AI agent: setup, tools, measuring with an agent
- FAQ
- Privacy and usage counts
- Changelog
The same pages are on the website, and llms.txt indexes them for AI assistants.
Pull requests are welcome: CONTRIBUTING.md has the setup,
the tests and what a PR needs, and docs/dev the
architecture. npm run dev:mock runs the whole
app with made-up measurements, so any machine will do.
Inspired by python-wifi-survey-heatmap. @hnykda started this version to get heat maps on a Mac with a browser UI; @richb-hanover built the four-tab interface and much of the platform support. The WebGL heat map renderer came in #35. An early screen recording shows the original single-page version.