Indie Hackers
I tried it on my site, and the H1 finding was accurate. Fixed all three. Good, specific finding. Not generic advice.
Free full report, no signup
Find what is slowing your site.
Find what is slowing your site.
Down to the file.
Site speed reports and uptime monitoring, in one place.
Paste a web address. See which file is slowing it. See what to change first.
- No signup
- We only read your site, nothing changes
Do not wait here. Keep this link.
Opens the report whenever it is ready, on any device. Nothing to sign up for.
Or email the report when it is ready
One email when it finishes, nothing else. Gmail may file the first one under Promotions.
2in the queue
You are second in line.
PafCore runs one scan at a time. Each one loads the page three times in a real browser, which one measuring server cannot do twice at once while this is the free public tier.
You do not have to wait here. Add your email under the link above, or copy that link and close the tab.
Starting
- Asking the server to start
- Loading the page, first run
- Loading it again, to see what moves
- Third run, then working out the causes
You can close this tab. The link above brings you back.
Your report
on a phone
Score looks too high? Check Lighthouse is set to Mobile. Desktop can be thirty points higher on the same page.
Watch this site
A score check on a schedule, and an email the moment the site stops answering. This report is one moment in time.
Start here
Send this to your developer, or copy the fix prompt into any coding agent.
Two halves to this. Make the change, then come back and press Check if it worked at the bottom. The score moves on its own, so on its own it cannot tell you.
Going to try this?
We will email you once, in three days, with a link to check whether the score really moved. One message, then nothing.
Already made the change?
We will measure your live site again and compare it with this report, against how much the page moves on its own. 2 credits.
Where you lost points
These overlap. Don't add them up.
What to try
Also worth a look
Not yours to fix
Optional next steps need credits
Everything above is free. Signing in gives you 5 credits to try the paid steps, then a pack if you want more.
Credits are paid: $5 for 20. Pro Watch is separate at $19/mo.
Quick question
Did it pick the right problem?
If it got your page wrong, say so. That helps more than a yes.
The full report, as the terminal prints it
That is the whole report. Everything below is about PafCore, not about your page.
0
pages measured
0
points the average page moves on its own
0
moved 10 points or more with nothing changed
Counted from every scan we have finished, not an estimate. This is why we load a page three times instead of once.
Public reports
See what other people found.
Open any report. Same free answer you get when you paste a site.
No public reports yet. Be the first.
One report per site. Anyone with the link can open it.
The problem
You get a long list. You still do not know what to change first.
Most of that list does not raise the score. We name the file that would.
5 of 49
Items in a typical speed report that add to the score. The other 44 are tips. Nothing says which tip would move one of the five.
0 of 25
What one real site already scored on the part its whole image fix list pointed at. Doing all of those fixes would have earned almost nothing.
9
Points between two checks of the same untouched website. Bigger than many real fixes are worth, which makes one score hard to trust.
The score moves even when you change nothing
measured, not an example
run one
60
run two
51
9 points apart
Same website. Ran it twice. Changed nothing. Still nine points apart. So a small rise after a fix can be luck. Every report shows how far your score moves on its own, so you know when a rise is real.
A long report is mostly tips that add zero.
Extra script, blocked images, big files. All useful to clean up. On their own they add zero to the score. They only help if fixing one speeds up one of the five that count, and most reports never say which.
Every square is one item in a speed report. The amber ones are the only ones that add marks.
5 scored · 44 not
You can work all afternoon and the score barely moves.
Going from 7.3 seconds to 6 seconds feels better for visitors, and it is worth about two points. The scale gets stingy on a slow page.
Marks out of 25 for how long the biggest thing takes to show.
22.5 marks at 2.5s · 1.3 marks at 7.3s
Uptime monitoring
A fast website is no use when it is down.
You find out the minute the site stops answering. Two pages watched free with an account.
shop.yoursite.com
checked 41 seconds ago
Not answering. The server refused the connection.
Email sent at 04:12. One more when it is back. Nothing else.
Every minute
A status code, or words that have to still be on the page, because a broken deploy still answers 200.
One message, not ten
Two bad checks before we say anything, so a single dropped connection does not wake you at four in the morning.
Before it expires
Warned 30, 14 and 7 days before a certificate runs out, and 60 days before the domain name does. Both take a site down for real.
A page you can share
One link for customers, 90 days of uptime on it. Plus heartbeats for jobs with no address, like a backup that never ran.
What you get
What you do with a report.
The free check does the first three. You only pay if you want us to test a change, or a file to download.
01
The file
What is slow, and which file is doing it.
02
What first
What to change today, and how many points it might earn.
03
After you change it
Check again. Did the score really move?
04
A clear answer
It worked, it did not, or we cannot tell yet.
From builders
They tried it on a real site.
They said it named the right file. Not a generic list.
- 3 builder notes
- IH Indie Hackers thread
Indie Hackers
Pinpointing the exact file that blocked the logo gives a clear place to start. The quiet gap is the best feature on the page. A small score jump after a deploy can just be noise.
Indie Hackers
Quiet gap is the most useful part. Clients stop chasing the number and ask about the file. Start here was clear, and the file detail beat a score alone.
Ways to use it
You do not need a terminal.
Paste a site here. A terminal, Cursor, or an API are there if you want them.
Website
Paste a site. Free report. Sign in to check a fix, find the file, or get a smaller image.
pafcore.siteTerminal
Full report from our servers. Sign in, scan, then see if a suggested change would help.
pafcore login
pafcore scan https://yoursite.com
pafcore test 1Cursor or Claude
MCP tools so an agent can scan, check a fix, and find the file.
pafcore mcp installAPI
Same routes the site and CLI use. Session header. Curl ready.
POST /v1/scans
GET /v1/findings/:id/:nNeed score and noise only on your own machine, with no file names?
That is a different free tool: npx @pafcore/cli scan.
See measure only CLI.
The tools that already exist
A score does not tell you what to change first.
Google already grades the page. You still leave with a list.
Lighthouse
Opens your website in a real browser and gives it a score. It has got better, and it can name files. It will tell you which image is too big.
What it leaves you It grades one run. Run it again and the number changes, and it will not say whether that change means anything.
PageSpeed Insights
Lighthouse on Google machines, plus real visitor data from Chrome. The visitor half is the part worth having, because search cares about it.
What it leaves you That half takes weeks to show a change you ship. Neither half tells you if this morning's fix beat the score moving by itself.
AI coding tools
Good at writing the fix. Point one at a report and it can shrink an image or split a big script. Faster than doing it by hand.
What it leaves you It reads the same report, so it picks the biggest number rather than the one you can win, and it checks its work by running once.
Questions
Questions people ask first.
Is this just Lighthouse with a nicer look?
It uses Lighthouse for the lab numbers, so the raw scores match. Two things are different. Lighthouse grades one run and stops. PafCore loads the page three times and shows how far the score moves on its own, so you can tell a real gain from a lucky run. It also finds the file behind the slow part, then orders fixes by how many points they might earn.
If you only need a grade, Lighthouse is free and already on your computer. Use it.
Can I paste the report into Cursor and let it fix things?
For the fix itself, often yes, and faster than you would. The hard part is choosing what to fix. The tool reads the same report you do, so it picks whatever number looks worst, not whatever you can win, and it checks its work by running the test once more.
One more run cannot tell a real gain from the score moving by itself. Measuring that move is where this starts.
What will it cost?
The full report on this page is free for public websites. Three free reports a day. After that, another report costs 1 credit when you are signed in. Fair use still stops one site soaking the scanner.
Signing in gives you 5 credits for the next steps: see if a suggested change would help, find the file, download a smaller image, or get code to paste. A pack is $5 for 20 credits. Credits also buy another full report after your free daily limit. Pro Watch ($19/mo) is separate: schedules, history, alerts, and 60 credits each month.
When can I use it?
Now, at the top of this page. Paste a website address and you get the real report: the score, how far it moves on its own, the file behind the slowest thing, and what to fix first.
Same engine also runs from the terminal, Cursor or Claude (MCP), and the HTTP API. See Ways to use it and the docs.
There is a separate free command for score and movement only on your machine. It does not name the slow file the way this page does. Details: measure only CLI.
Can I use it from Cursor or the terminal?
My score went up but Search Console has not changed. Why?
They measure different things on different clocks. PafCore is a lab score: one page, loaded now, on a simulated slow phone. Search Console shows field data from real Chrome users, averaged over the last 28 days.
A fix you shipped today cannot move a 28 day average today. Expect the lab number to move straight away and the field number to follow over the next few weeks, and only if real visitors actually hit the page you changed.
What exactly do you measure?
Headless Chrome through Lighthouse, performance only, three separate loads of the one address you gave us.
Mobile by default: a 412 by 823 screen, Lighthouse's own Slow 4G settings at about 1.6 Mbps, and the processor slowed to match a mid range phone. Throttling is simulated, the same method Lighthouse and PageSpeed Insights use, so the number is comparable to theirs.
Every load starts with an empty cache in a fresh profile, so nothing carries over between runs. We only read the page. Nothing on your site is changed.
Does PafCore do uptime monitoring as well?
Yes. We ask your page for an answer as often as every minute and email you when it stops giving one, and again when it comes back. It waits for two bad checks first, so one dropped connection does not wake you.
A free account watches two pages every five minutes. Pro watches twenty, and adds SSL and domain expiry warnings, heartbeats for jobs with no address, a status page you can share, and quiet windows so a planned deploy does not alert.
Will it warn me before my SSL certificate expires?
Yes, on Pro. We read the certificate off your site once a day and warn at 30, 14 and 7 days.
We read the domain registration too, and warn at 60 days. A lapsed domain takes the site and the email with it, and unlike a certificate it usually does not renew on its own.
Does it work on any site?
Anything a browser can open without signing in, to start with. Sites behind a login, and anything slow only after a click, are harder and come later.
It will not work on everything. When it cannot measure something, it says so rather than guessing.
Who builds it?
A small product, shipped in the open. Decisions and progress are posted on X. Security and product mail: hello@pafcore.site.
Scoring guide
Get the scoring guide.
How the speed score is really worked out. Not three sentences and not only the source code. Read it here, or leave an address and we will send it plus a note when something new ships.