Website security · part of Studio
Website security, checked every hour and fixed by our engineers.
Attackers now use AI to turn a published fix into a working attack, sometimes within a day. We check your website every hour, and again within five minutes of any change, for known flaws, poisoned packages, exposed files and leaked keys. When something is vital, our engineers are alerted at once and fix it. You get the fix, not the alarm.
- Every hour
- Within 5 minutes of any change
- Engineers alerted at once
Works with
- Any website whose code is in GitHub, whether we built it or not
Every hour, Studio reads the site's lockfiles and looks up every package and exact version. It also checks the certificate and the security headers, and once a day whether /.env or /.git/config opens.
When anyone publishes or merges, the site is checked again within five minutes, so a new package never waits for the next hour.
A critical flaw in a package that reaches the live site, a flaw on CISA's list, or a version known to contain malicious code is vital. Our engineers get one email at once, whichever sites it affects.
Studio moves the package to its fixed version on a separate branch of the site's code, then checks that branch again to confirm what it clears.
Nothing merges by itself. An engineer approves the exact change Studio checked, and every approval is recorded.
Your team sees: Security update applied: next
Example morning
- 22
- zero-day exploits in August 2026, up from 8 a month in 2025
- 2 days
- from React2Shell's disclosure to attacks on Next.js sites
- 868
- npm packages poisoned by one worm, ChainDrop, in August 2026
- 1 hour
- the longest a new advisory waits before Studio checks your site against it
Sources: Google Threat Intelligence Group (November 2026), Rapid7, Aikido. Links in our post on website security monitoring.
Why now
The gap between a fix and an attack is now about a day.
In November 2026, Google's Threat Intelligence Group reported an attacker's AI framework that reads each new software patch, works out the flaw it fixes, writes an exploit and runs it, with no person involved. Its targets were websites.
So the moment a fix is published, every website that hasn't updated is on a list. A site updated once a quarter, or when someone remembers, is exposed for most of that time. These are the attacks websites are facing:
- Known flaws in the packages a site is built from, like React2Shell in Next.js
- Poisoned packages: hijacked releases that steal the keys on any machine that installs them
- Exposed .env and .git files, and keys committed to the code
- Attacks on the AI coding assistants that now change websites
- Missing security headers, and certificates that lapse
What we check
Inside the code, and from outside, every hour.
Six checks cover the attacks above. They only read: your live site doesn't change until an engineer approves a fix.
Packages
Known flaws in what your site is built from
Every package and exact version in your site's lockfiles, looked up in the open vulnerability database, which includes GitHub's advisories. Your site's own packages are also checked for newer releases.
Used in attacks
Flaws attackers already use
Every match is checked against CISA's list of flaws known to be used in attacks, read twice a day. A flaw on that list in a package that reaches your live site is vital.
Supply chain
Poisoned packages
A version known to contain malicious code is vital even if it only builds your site, because its code runs on every machine that installs it. Studio's own fixes never install a package or run its code.
From outside
Certificate and security headers
When your security certificate ends, the headers that tell browsers to use https, run only approved scripts and refuse to frame your pages, and whether your site announces what it runs on.
Exposed files
Files that should never open
Once a day, whether /.env or /.git/config can be opened from the web. The first holds passwords and keys. The second opens the way to your code and its history.
Keys
Keys committed to the code
A committed .env file with real values, or a private key file, in your site's repository. The values themselves are never copied into a report.
How it works
From a published flaw to a fixed site, without anyone on your side lifting a finger.
Checked every hour
And within five minutes of any change going live, and whenever our engineers ask. A new advisory for something you use is caught by the next hourly check.
Ranked by what matters
Vital means a critical flaw, or one used in attacks, in a package that reaches your live site, malicious code anywhere, an open file or a certificate about to lapse. The rest is ranked for an engineer's list.
Engineers alerted at once
Anything vital reaches our engineers by email straight away, as one message however many sites it affects. Everything else is in their summary each morning.
Fix prepared on a branch
Studio updates the package list on a separate branch of your code. It installs nothing and runs no package's code. Anything it can't prepare, such as a new major version, goes to an engineer.
Approved, then live
An engineer approves the exact change Studio checked. Nothing merges by itself, every approval is recorded, and the site is checked again to confirm the fix.
What you see
You get the fix, not the alarm.
You are not sent alarms, and nobody on your team has to read a security advisory. Your website's page in Studio shows that it is watched and when it was last checked. Once something is fixed, a one-line note says what was done.
Your AI assistant can tell you the same, in Claude or Codex.
Checked every hour and within five minutes of any change. You don't need to do anything.
A one-line note appears once a fix is live. Several fixes of one kind on one day are one line.
Your AI assistant can tell you the same. Studio tells it to report what was done and not to guess.
Ask in Claude or Codex: What security updates were applied this month?
Example
AI and your website
The AI assistant drafts. People decide.
Attackers now go after the AI coding assistants that change websites. Studio, the service the security watch is part of, lets your team change the website from Claude or Codex, and it is built so that an assistant never publishes on its own:
- Every change is previewed, and nothing goes live until a person with the right role approves it
- Each person has a role: who can draft, who can approve and who can publish
- Studio tells your AI assistant to treat words on pages, in search terms and in comments as content, never as instructions
- Studio's GitHub app can read and write only the contents of the repositories you give it
- Every action is recorded, and a published change can be undone
What it doesn't replace
Fast on known problems. Not a penetration test.
Monitoring finds known problems quickly and ends in a fix. It has limits, and we would rather say so:
- It doesn't test your site's own code, your hosting account or your domain settings
- It can't see flaws nobody has published yet, or a poisoned release in the hours before the databases list it
- A match between a version and an advisory isn't proof your site can be attacked: an engineer decides
- Today it reads JavaScript and PHP lockfiles. Sites built with other languages are checked from outside
Getting started
Have your website watched.
What you need
- A website whose code is in GitHub, whether we built it or not (we check this for you)
- Access for Studio's GitHub app to the site's repository
What we do
- Connect your site to Studio and run the first check within minutes
- Check it every hour, and within five minutes of every change, from then on
- Alert our engineers to anything vital, and prepare and approve the fix
- Connect Studio to Claude or Codex for your team, if you want to make changes too
Security comes with every website in Studio. The price depends on how many websites you run. Tell us about yours and we will send it.
Tell us about your website.
Free website audit
Not sure where your site stands? Start with a free audit.
Six checks, security included, alongside speed, search, AI visibility, accessibility and content. We run it for you and email the report.
Where should we send it?
Questions
Questions about website security.
What do your website security services include?
Every website in Studio is checked every hour and within five minutes of any change: the packages it is built from, against the open vulnerability database and CISA's list of flaws used in attacks; its certificate, security headers and exposed files, from outside; and keys committed to its code. Our engineers are alerted to anything vital, and they prepare and approve the fix.
How is this different from a website security scanner?
A scanner tells you what is wrong and leaves the rest to you. Here the checks run every hour by themselves, our engineers are alerted, and the fix is prepared, approved and published for you. You see what was done.
How quickly is a new security flaw picked up?
A new advisory for something your site uses is caught by the next hourly check, so within an hour of it reaching the open databases. A change to your site is checked within five minutes of going live.
Are we protected from supply chain attacks?
Studio treats any package version known to contain malicious code as vital, even one that only builds your site, and our engineers move you off it. It can't catch a poisoned release in the hours before the databases list it, and it doesn't watch the computers or build systems that installed it.
Will we get security alerts?
No. Alerts go to our engineers, who fix the problem. You see that your site is watched, when it was last checked, and a short note of each fix once it is done.
Does it replace a penetration test?
No. It finds known problems quickly. It doesn't test your site's own code, your hosting account or your domain settings, or find flaws nobody has published yet.
Does it work with websites Agitech didn't build?
Yes, with any website whose code is in GitHub. Today it reads JavaScript and PHP lockfiles, and it checks every site from outside, whatever it is built with.
Do you check for exposed .env files and leaked keys?
Yes. Every day Studio checks from outside whether your site serves /.env or /.git/config. It also checks your code for committed key files and .env files with real values. Our engineers are alerted to anything it finds and handle the fix.
Do you secure Next.js and React websites?
Yes. Every hour Studio reads your site's JavaScript lockfile and checks every package, including Next.js and React, against the open vulnerability database and CISA's list of flaws used in attacks. When an update fixes a flaw, our engineers prepare it on a separate branch and publish it once it is approved.
Is this a web application firewall?
No. A firewall, such as Cloudflare's, filters traffic before it reaches your site. Studio keeps the site itself free of known flaws by finding them and fixing them. They do different jobs, and many sites use both.
What leaves Studio when it checks our site?
Only package names and versions, sent to public databases to look them up. Nothing about you, your site or your code.
How much does it cost?
Security comes with every website in Studio. The price depends on how many websites you run: tell us about yours and we will send it. You can also start with a free website audit.
Find out what your website is built from, and what is open.
You send your website
Tell us the address and who looks after it today.
We check it can connect
Any website whose code is in GitHub, whether we built it or not.
You decide
We send a price. Go ahead or walk away. There is no obligation.