The time from a vulnerability’s disclosure to exploitation has gone from 32 days in 2022 to mere hours in the post-Mythos era. CISOs we’ve spoken to report 10,000 to 20,000 commits a week, tested by security teams they don’t have the budget or time to grow. At the same time, they report needing to know whether a release exposed them within 24 hours or it’s simply too late. 

With attack surfaces infinitely growing and evolving, especially as companies increasingly integrate MCPs and AI agents alongside a plethora of other features, security teams struggle to keep up. Manual penetration tests cannot keep the pace of software generated code. Legacy scanners fail to find any meaningful issue. AI-powered pentesting companies focus on point-in-time assessments while software is built continuously. 

Escape Continuous AI Pentesting closes the gap between the outnumbered security team, their ever-expanding attack surface as developers ship code with AI, and their ever-shrinking time-to-remediate as attackers can now exploit vulnerabilities within a day.

By only pentesting what each release changes, continuous pentesting runs on every merge, deploy, or a schedule you choose. This gives modern security teams the pentest-depth testing they need on all of their web applications, APIs, and MCP servers, that keeps pace with how often their engineers ship, and how fast attackers move. 

Companies’ attack surface is growing faster than they can test it. AI-written code, new applications and agents with real permissions are all multiplying, and attackers are using the same models to move faster. By 2027, I expect AI-driven attackers to autonomously discover and exploit previously unknown complex vulnerabilities within minutes, without relying on public disclosures or human-written exploits. - Antoine Carossio, CTO and Co-Founder of Escape

Here’s what security teams get:

• Near-real time pentest results for all your engineers:  Our whitebox integration reads your source code for changes, works out which ones matter, and runs a pentest only on the diff and wider areas of the app it affects. This allows us to give your engineers high quality pentests results in minutes instead of days.
• High quality findings that matter to your business: Instead of starting everything from scratch, Escape ingests your past pentest reports, bug bounty findings and other context to give you only results worth fixing instead of growing your backlog with low priority items.
• One offering for your compliance evidence: Provide the maximum level of evidence in a single scan. Continuous Pentesting provides evidence for dynamic scanning, penetration tests and more. Human made OSCP certified pentests reports are available to meet even the highest compliance requirements. 
• No duplicate backlog: Known issues come back as replays - not new tickets - and every open issue is re-checked on every run. So fixes are automatically verified and known issues don’t grow your backlog.

Get pentest results in near-real time

A point-in-time pentest is already outdated by the time you’ve downloaded the PDF. These are simply insufficient to give security teams an actual insight into their vulnerability exposure. The solutions were either to wait weeks, even months, for the next pentest and accept the gap or rerun the entire AI pentest - up to 24 hours across an app - to see the difference that a handful of changes made. And when the number of zero-days among CVEs has risen from 16.1% of CVEs in 2018 to 67.2% in 2026, that waiting period is simply too long. 

They'd send us back a spreadsheet, we'd fix whatever there was, uh, and then somehow we'd get another spreadsheet the next year that was like even longer, and it's not the greatest, to have to wait like a whole year to find out if we're doing a good job or not. - Kevin Richards, Senior Application Security Manager, Veg ER for Pets 

That changes now. Continuous AI pentesting is the part of Escape AI Pentesting that pentests just what each release changes. On every merge, deploy, or on a schedule of your choice, our whitebox integration scans your source code for changes, decides what’s actually effected a meaningful change, and runs a pentest only on the diff. That takes a scan from 78 minutes to 4.9 minutes in our benchmark. 

Before opening our MCP server to customers, we wanted it tested the way AI-driven attackers would approach it. Escape’s AI Pentesting fit into our pre-launch process and gave our dev team clear, actionable results we could act on quickly.” Jun Carreon, Principal Information Security Analyst, Arkose Labs

Decrease your exposure time without increasing your noise

Open issues are re-checked on every run and known issues aren’t made duplicates

One of the biggest concerns is pentesting more often will mean more noise added to an already insurmountable backlog. Escape’s Continuous Pentesting set out to solve this precise problem, and we engineered it specifically for high signal-to-noise ratio.

How?

• Every open issue is re-checked on every run and kept as one issue, with our deduplication agent
• A rerun repeats every open finding, whereas other continuous only tools retest the fixes you flag
• Escape focuses every re-run at your precise exposure risks by learning from past bug bounty and pentest reports

What truly differentiates a continuous pentesting solution from one that just deploys more pentests is something designed to streamline your backlog, not snowball it. In our benchmark, diff runs finished 16x faster than a full assessment, caught 91% of the vulnerabilities introduced by each change.

So with these key features, Escape Continuous AI Pentesting is able to achieve that signal-to-noise value for security teams so they can spend time fixing findings, instead of validating them.

Remediate vulnerabilities as soon as they surface, instead of losing them to an unopened PDF

With an annual pentest, critical issues can get lost in a graveyard of PDFs, with no clear owner, remediation guidance, or timelines. So they simply don’t get fixed and are sat, exploitable, until the next annual review. 

Continuous AI Pentesting gives you visibility on these issues, routes them to their owners, and comes with tailored remediation guidance (or a prompt ready for the model of your choice), as soon as the introduced vulnerability is deployed. Constant visibility means your engineers don’t need to trawl through code they wrote months ago to try and make sense of it, but can immediately implement fixes and eradicate your exposure.

How Escape prioritizes meaningful findings

So what does this mean? All of your fixes are automatically verified, known issues don’t grow your backlog with duplicates, and by learning from your application’s context, past bug bounty reports, and any other context insights you provide, it only surfaces findings that are actually meaningful to your business. 

By importing past pentest reports, bug bounty findings, and any other context documents, Escape can turn them into attack scenarios specific to your organization. If five of your past reports were XSS, your run gets tailored XSS scenarios. Continuous runs play those scenarios against every route a change reaches, so the agents spend their hour on the weaknesses your applications have actually had. Better context means more findings worth fixing and fewer false positives to triage. The scenarios apply across every profile in your organization, so one team’s past finding sharpens testing on everyone’s apps.

Through building tailored attack scenarios by learning from exactly what was critical to your organization in past pentest and bug bounty reports, Escape Continuous Pentesting can match the precision and quality of findings of these forms of vulnerability discovery, in a fraction of the time. It’s the only automated testing solution on the market that’s able to provide these tailored findings continuously, which means 

Total transparency with reports for your auditors and your engineers in one click

For highly regulated industries (banking, insurance, fintech, healthcare), a finding is only half the job. The other half is proving what was tested and how to a wide range of stakeholders across the company and external to it. The other biggest problem with increasing the frequency of AI pentests is increasing your surface area to the dreaded back-and-forth with engineering on what’s actually an exploitable finding. 

Reasoning logs come with a complete trail of what happened

Escape’s AI pentesting has a complete trail of exactly what was tested and why in our reasoning logs, so you can show stakeholders precisely what happened, and hand developers the reproduction path so they can get to fixing, not flagging false positives. While a point-in-time pentest report also gives you this, a continuous run gives you the logs for every change. That’s  crucial for frameworks like PCI DSS v4.0.1 which requires pentesting at least once a year and after any significant application change. 

As one of our existing customers Bridgetech found, they were able to use their pentest reports for both customers and auditors to prove their security and diligence.

The other thing is it allows us to demonstrate that we're diligent to existing and new customers. So, if customers will often ask for a penetration test against our platforms, and some customers will accept the output of tools like Escape. That is massively beneficial for us.” — Ben Dalby,  Chief Delivery Officer, Bridgetech Group

Escape can give you pentest-depth testing on every release, continuous scanning for known issues on every build, and a human-verified OSCP sign-off for auditors, satisfying the needs of most major compliance frameworks without having to switch platform for every test. 

Complete compliance all in one place

Paired with fully scoped pentests and OSCP certified tests, you get all of the compliance evidence you need in one solution. For compliance frameworks like PCI DSS, instead of scoping a test every time you make a major application upgrade, continuous pentesting with human-verified certification ensures you’re always audit-ready without adding that as another task to the backlog.

Runs follow how you release: Here’s how it works

You choose one trigger per profile:

• On every merge. The Escape GitHub App sees a push to the branch you watch and starts a run. There's no workflow to add.
• After a deploy. Your CI calls Escape's API once your deploy is done, so the run tests what's live. The product gives you a ready GitHub Actions workflow; any other CI makes the same CLI call. 
• On a schedule. Daily, weekly, or monthly.

Continuous means testing at the pace you release, but you can decide which flow makes the most sense for your organization, tying to a production schedule or ensuring every new piece of merged code is checked. Only one run happens at a time per application. A merge that lands while a run is in progress doesn't start a second one: the next run covers everything merged since the last completed run. Runs happen on Escape's side, so your build never waits on them.

How a run scopes itself

Each run reads the commits since the last run on each watched repository, one branch each. Then it:

1. Analyzes the changed code and reports the issues the change introduced.
2. Maps the change to every route it reaches, up to 20 per run. A change to a middleware or an auth helper opens risk on routes whose code never moved; diff and pull request tools only look at the changed files.
3. Attacks those routes on the deployed app with the agents of Cascade Iguazu, Escape's multi-agent AI pentesting engine. When a route fronts a chatbot, an AI agent, or an MCP server, the run applies the techniques of AI Red Teaming, included in AI Pentesting.
4. Replays every open issue against the live app, in parallel. An issue closes only when its fix is verified, and an issue you've marked yourself (ignored, false positive) keeps your status.

It reads Go, Python, TypeScript and JavaScript, Java, Ruby, PHP, C#, Rust, C and C++, Kotlin, Swift, and Scala. 

Get started: test every commit, with the power of Cascade behind it

With continuous pentesting, the deeper exploit engagements that were once reserved for annual compliance requirements can now be run at the same pace as the deluge of code being written and the onslaught of AI-powered attacks. 

As teams ship fast and attackers exploit even faster with AI, security teams can now match that speed with pentest-grade coverage. Deeper coverage, faster, and all ameliorating your auditability and without increasing your noise. Escape’s Continuous Pentesting equips the outnumbered security team with the arsenal they need to come out ahead in a post-Mythos security era, and meet compliance requirements, all in one solution. 

Complementing our release of AI Red Teaming, Cascade Iguazu, and a plethora of other features, continuous pentesting is another milestone for Escape, with even more yet to come.

Ready to make the move to continuous? Book a demo today and we'll show you how Escape can close your exposure window.