hint: one linting engine, three front ends, and a release list that is not a release list
💡 A hinting engine for the web
At a glance
- What is it?
- A configurable linting tool that checks pages for accessibility, speed, and cross-browser problems, shipped as a command line tool, a browser extension, and an editor extension from a single monorepo. Its tags tell you almost nothing about its history, which is the first thing to understand about dating it.
- Who is it for?
- hint fits a team that wants accessibility and compatibility feedback on real pages rather than on source files, and that wants the same rules in continuous integration, in the browser, and in the editor. Two things to check before adopting it.
- Can I use it commercially?
- Yes. Apache-2.0 is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 123 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 21, 2026, and from our analysis. They are not legal advice.
Star history
- webhintio/hint3,701 ★ · -2
| Repository | 2026-09-24 | 2026-11-21 |
|---|---|---|
| webhintio/hint | 3.7k | 3.7k |
-2 Stars in the last 58 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
Three ways to run the same checks
The tool is presented as configurable rather than opinionated: you choose what it checks, and the defaults are a starting point rather than the product. What it looks at is named in one sentence, covering accessibility, speed, cross-browser compatibility, and the general category of best practices and common errors. The delivery is the more interesting part, because the same engine appears in three places. There is a command line tool, which is the one you wire into continuous integration. There is a browser extension, which is the one a person runs against a page they are looking at. And there is an editor extension, which is the one that fires while you work. Sharing one engine across all three is why the repository is a monorepo, and it is also the reason the configuration has to be portable between a shell, a browser panel, and an editor panel.
One command for a quick look, a script task for the long term
The quick start is deliberately frictionless: with a current JavaScript runtime installed, a single command analyses a URL with the default configuration and needs nothing else on the machine. That path is for trying the tool, not for adopting it. The documented install is a development dependency of your own project, plus one entry in your package manifest so the command has a name.
npm install hint --save-devThe manifest entry is a scripts key that points at the binary, and the run command then takes the URL after a separator. There is a shortcut for people using the other package manager: it can call the binary directly with no task at all, which is convenient locally and unhelpful in continuous integration, where you want the script task so the invocation is recorded somewhere visible. Two requirements are named along the way, the runtime floor and the runner used for one-off testing.
The documentation exists twice, and the newer copy is local
There are two sets of documentation for the same tool and the readme tells you which to trust. There is an online user guide, and there is a version of the same guide inside the repository, and the sentence covering both says to look at the local version for the most recent content. That is a small admission with practical consequences: if you are reading the published site and something does not match the installed version, the repository copy is the one to compare against. The contributor documentation is split the same way, with an online version and a local one, and there is a third split for the two extensions, each of which carries its own contribution file inside its package rather than pointing at a shared page.
The tags are packages and a branch, not versions
This is the thing to know before you try to date the project. The recent release list contains three entries and none of them is a product release. One is a branch named for distribution files, tagged as if it were a release. The other two are individual parser packages at a version in the low twos, both published within minutes of each other. Meanwhile the default branch was pushed to in July 2026. Read together, that says the tags are per-package artefacts and that nobody publishes a version for the tool as a whole, which is consistent with a monorepo that ships each flavour separately but leaves no single version to point a dependency at.
Each flavour is its own package with its own instructions
The contributing section states the layout plainly: this project follows a monorepo pattern, which means the code for every flavour, including the command line tool, the browser and editor extensions, the checks themselves, and the formatters, all lives here and is published as separate packages. Building from source needs a recent runtime and the other package manager, then two commands at the root, an install followed by a build. The build is followed by an apology for how long it takes, which is the kind of warning that only appears in a monorepo where a full build genuinely runs the compiler many times. The two extensions do not share the top-level contribution guide; each has its own file inside its package, which is how you learn the difference between the three front ends without reading all of them.
A handful of dependencies are pinned hard, and one pair disagrees
The root manifest has no dependencies of its own because it is a private workspace root, and everything it lists is for building the project rather than running it. What is interesting is the resolutions block, which forces a few transitive packages to older lines: a JSON schema library at a version below one, a fetch implementation held at major two, and a package called for chaining callbacks. It also pins the browser compatibility dataset, which is the dependency that makes cross-browser checks possible and therefore the one worth being careful about. One inconsistency is visible in the same list: the typescript linting plugin is on its fifth major while its parser is on its fourth, which is a combination that usually works and occasionally does not.
The build regenerates references, then hands over to a custom driver
The root build script is three steps with a name that explains the first: it cleans, then updates references, then runs a script from the repository's own scripts directory to build or test everything. That middle step is the one to notice, because it means generated cross-references are not committed by hand, and a change that forgets to regenerate will be caught by the build rather than by review. TypeScript is built with project references, and there is a separate script that copies the non-TypeScript files into the distribution, so the output is assembled from compiled code plus copied assets. The remaining root entries tell you what else the project considers its own: a coverage configuration, a markdown linter, a markdown link validator, a release directory, an editor directory, a shared types directory, and two sets of pipeline files for the hosted build service.
Editorial conclusion
hint fits a team that wants accessibility and compatibility feedback on real pages rather than on source files, and that wants the same rules in continuous integration, in the browser, and in the editor. Two things to check before adopting it. Read the release history carefully, because the recent tags are packages and a branch rather than product versions, so you cannot date the tool from them. And decide early which front end is canonical for you, because the documentation admits the local copy is ahead of the published site.
Frequently asked questions
What is webhint used for?
It is a configurable linting tool that checks pages for accessibility, speed, cross-browser compatibility, and common errors. The same engine runs as a command line tool, a browser extension, and a VS Code extension, so one configuration covers all three.
How do I run webhint without installing it?
With the runner, against a URL, using the default configuration. The documented command analyses the page and prints its findings without needing anything installed beyond a current JavaScript runtime, which makes it a reasonable first check before you add it to a project.
How do I add webhint to my project?
Install it as a development dependency, add a scripts entry that points at the binary, and run the script with the URL after a separator. If you use the other package manager you can call the binary directly without adding a script task at all, which is convenient locally and less visible in continuous integration.
How current is the webhint documentation?
There are two copies, an online user guide and one inside the repository, and the readme points you at the local copy for the most recent content. Each extension also has its own contribution file inside its package rather than sharing the top-level guide.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/webhintio-hint)Star history chart for your README
[](https://hysenlabs.com/projects/webhintio-hint)Was this useful?