Unlighthouse
It is an open tool that runs the browser's audit across all of a site's pages, discovered automatically, and gathers the results into a single dashboard. It fills the gap between a page-by-page audit and a crawler, without replacing either.
What it does
It discovers a site’s addresses — through the robots.txt file, the sitemap and internal links —
then runs the browser’s audit engine on each of them, in parallel, and gathers everything into a
single dashboard. One command is enough to set it off, with no prior configuration and no account to
create.
The tool is published under the MIT licence, declared in the public repository hosting its code.
officialSource : Unlighthouse, public repository and declared licence
Development is under way: the public repository carries changes dated July 2026, a few days before the check.
officialSource : Unlighthouse, repository commit history
This entry exists for a precise reason, and it is structural rather than functional: the Lighthouse entry criticises the browser’s audit for judging one page at a time, when the faults that cost money are faults of distribution. That is exactly the gap this tool fills, with the same audit engine and at no cost.
What it does not do
Auditing every page and inventorying a site are two distinct operations: the first judges pages one by one against a list of checks, the second establishes the complete list of addresses and their responses.
hypothesis
It does not hand back the raw inventory — the complete list of addresses with their response codes, their redirects and their inbound links — which remains what a crawler does best and what its licence actually buys. It judges pages; it does not draw up a survey.
A tool auditing a whole site samples addresses that resemble one another, failing which the run would take an impractical amount of time: the report then describes families of pages, and not each of them.
hypothesis
That point deserves understanding before drawing conclusions from a report: on a site where thousands of addresses share the same shape, the tool keeps a few of them as representative. The report then describes families of pages. That is the right compromise for ranking a piece of work, and the wrong tool for proving that no page has a problem.
Like any tool of its kind, it says nothing about what the engine kept of your site — that is read in its console — nor about what your visitors actually experience.
What it costs
Nothing: a permissive licence, no paid hosted offer announced by the author. The only entry cost is an up-to-date development environment on the machine running the command, which in practice rules out people who do not work with those tools — and that is a real barrier, even if it appears on no pricing grid.
When the paid tool stays the right choice
When you need the inventory, not the judgement. To prove after a migration that no old address returns an error, you need the exhaustive list of addresses and their responses: that is the job of Screaming Frog SEO Spider, and the sampling described above explicitly disqualifies this tool for that use.
The complementarity is clear, and that is rare enough to be worth saying: the crawler establishes what exists, this one ranks what has to be fixed.
Auditing a whole site before deciding which page to start with
hypothesis
- Effort
- moderate
- Time to effect
- immediate
- Risk
- none
- Durability
- lasting
Affiliation
This site takes part in no affiliate programme for this tool, and none exists: an open project pays nobody to be recommended. This entry earns nothing, and the order of this section has never depended on payment.
What does not work
A complete audit of every page on your site, free: no more need for a paid crawler.
Sources
- Unlighthouse — public repository, licence and history — Harlan Wiltonaccessed on
- About Search Console — Google Search Consoleaccessed on
Block revalidated on
Entries in this category
Frequently asked questions
- How does it differ from the audit built into the browser?
- By scale, only. It runs the same audit engine, but on every page discovered instead of one, then gathers the results. The question moves from “why is this page slow” to “how many pages on the site are”.
- Does it replace a paid crawler?
- Only partly. It walks the site and judges each page against a list of checks, but it does not hand back the raw inventory of addresses and their responses, which remains what a crawler does best. The two overlap without substituting for each other.
- Is every page really audited?
- No, and the tool says so: it samples addresses of the same shape so the run stays practical. The report therefore describes families of pages. That is a reasonable compromise, provided the sample is not read as an exhaustive inventory.
- Does it have to be installed?
- It runs from one command, with no configuration and no account to create, which makes it one of the fastest tools in this directory to try. It does assume an up-to-date development environment on the machine, which is not available to everyone.
- Do its scores say anything about ranking?
- No more than those of the single-page audit whose engine it reuses. They are scores calculated by the tool by its own weighting, and no engine publishes a per-page score that position depends on. They serve to rank a piece of work, not to predict a result.