Pentest Retest: Confirming Fixes Work, With Evidence
A pentest retest is a targeted re-run of the original attack against each finding you fixed, done to confirm the exploit no longer works and that close variants fail too. A credible retest produces evidence per finding: the request sent, the response received, a timestamp, the build tested and a clear status such as fixed, partially fixed or not fixed.
Continuous AI-powered penetration testing by Penetrify - find and fix vulnerabilities before attackers do.
What a pentest retest covers, and what it does not
A pentest retest (also called remediation verification or fix validation) is narrower than the original test. The tester takes each finding from the report, repeats the documented attack against the fixed build and records whether it still works. A good retest also tries close variants: a different encoding of the payload, the same flaw on a sibling endpoint, the same request sent as a different user role.
It is not a new penetration test. Code that shipped after the original engagement is out of scope unless you agree otherwise, so a clean retest says "these findings are closed", not "the application is secure". Keep that distinction when you hand the result to an auditor or a customer.
Three things often get passed off as a retest and should not be:
- The developer's own check. The person who wrote the fix tests the case they thought of. Verification should come from whoever found the issue, or from an equally independent party.
- A scanner re-run on its own. A signature scanner can confirm a missing header or an outdated library version. It usually cannot confirm that a broken access control or business logic flaw is closed.
- A ticket status. "Done" in the tracker means code was merged. It does not mean the exploit fails in production.
Evidence a verified retest should produce
A line reading "retested: fixed" is weak evidence. If you need to prove to someone else that a fix works, ask for one record per finding that contains:
- The finding ID and title from the original report, so the two documents map one to one.
- The date and time of the retest, with time zone.
- What was tested: the environment URL plus a build identifier such as a commit SHA, image tag or release version.
- The exact request or steps replayed, and the response received (status code, the relevant part of the body, or a screenshot).
- The variants attempted beyond the original proof of concept.
- The account or role used, for anything that involves authorization.
- A status, and who or what performed the retest.
Store the record next to the original report and the pull request that contains the fix. Original finding, code change, timestamped retest result: that chain is what keeps the remediation auditable months later, when nobody remembers the details.
Test every release, not once a year
Penetrify runs an autonomous penetration test of your web app and APIs from your pipeline and returns findings with fix guidance in minutes.
Start a scan →Retest statuses and closing rules
Agree on the vocabulary before the retest starts. Five statuses cover almost every case:
- Fixed. The original attack and its variants fail.
- Partially fixed. The reported payload fails but a variant works, or the fix covers some affected endpoints and not others.
- Not fixed. The original attack still works.
- Risk accepted. No fix was attempted. A named person signed off, with a reason and a review date.
- Not retested. The tester could not reach the feature, the environment was down or the fix was not deployed. This is not a pass.
Two rules keep the process honest. Close a ticket only on "Fixed", with the evidence attached. Keep a "Partially fixed" finding open at its original severity until the tester re-rates it, because a bypass of a weak fix is often as serious as the original issue.
Which kinds of solutions offer retesting after fixes
Traditional penetration testing firms
Many consultancies include one retest in the engagement, usually within a fixed window after the report is delivered. Others bill it as extra days. Confirm before you sign: how many retest rounds are included, how long the window stays open, and whether you receive an updated report or a separate attestation letter. The weak point is scheduling, since the retest depends on when the tester is next free.
Pentest-as-a-service platforms
These pair human testers with a web platform. You typically request a retest per finding from a dashboard and see the status change there, which suits teams that fix issues one at a time. Check whether retests are unlimited or metered.
Autonomous AI penetration testing
Tools in this category re-run exploitation attempts on demand, so a retest does not wait for a person's calendar. Penetrify is one example: it tests web applications and APIs by attempting exploitation, includes free targeted re-tests of fixed findings, and can be started from a CI/CD pipeline so verification runs when the fix deploys. That fits a startup shipping daily. It does not come with a certified human tester's signature, so if a contract names manual third-party testing, the retest letter has to come from that firm. The trade-offs are laid out in the comparison with manual penetration testing.
Bug bounty programs
The researcher who reported the issue is often asked to confirm the fix, and some platforms let you pay for that retest. Coverage depends on the researcher being available and interested, so do not rely on it for a deadline.
Vulnerability scanners
Scanners are the right tool for verifying patch-level and configuration findings across many hosts, and they record a scan date. They are the wrong tool for confirming fixes to findings a human or an exploit chain produced.
Getting a fix through retest on the first attempt
Failed retests are usually caused by narrow fixes. Before you ask for verification:
- Fix the class, not the payload. If the report shows one IDOR on
/api/invoices/{id}, check every endpoint that takes an object ID. Blocking the exact string from the proof of concept is not a fix. - Write a regression test that reproduces the exploit. It should fail on the old code and pass on the new. This also stops the issue returning after the next refactor.
- Deploy to the environment the tester will use and note the build identifier. Retests fail surprisingly often because the fix is merged and not yet deployed.
- Tell the tester what changed. One or two sentences per finding ("authorization check moved into middleware, applies to all
/api/*routes") lets them aim variants at the right place. - Keep the test accounts alive. Access control findings need the same roles and data that were used originally.
Timing the retest before an audit
Some frameworks require the retest explicitly. PCI DSS requirement 11.4.4 says exploitable vulnerabilities found during penetration testing must be corrected and the testing repeated to verify the corrections, so for PCI DSS penetration testing the retest is part of the requirement, not an extra. SOC 2 does not prescribe a penetration test at all, but an auditor who reviews one will generally ask how its findings were tracked to closure. More detail is in the guide to SOC 2 penetration testing.
Practical timing rules:
- Book the retest date when you book the test, and work back from the audit date to set fix deadlines.
- Retest critical and high findings as each fix deploys instead of batching everything into the final week. A failed retest then still leaves time for a second attempt.
- If the retest window is about to close and some fixes are not ready, retest what is ready and document the rest with an owner and a dated plan. An open finding with a plan reads better to an auditor than a finding nobody verified.
Frequently Asked Questions
What is a pentest retest?
It is a follow-up check in which the tester repeats the attack for each reported finding against the fixed build and records whether it still works. It covers only the findings from the original report, plus close variants of them. It is not a new penetration test, so code released since the original engagement is not assessed unless you agree to extend the scope.
What kinds of solutions offer verified retests with timestamped evidence and reporting?
Four kinds do: traditional pentest firms (often one retest inside a fixed window), pentest-as-a-service platforms (retest requests per finding in a dashboard), autonomous AI penetration testing tools (retests on demand, including from CI/CD) and bug bounty platforms (retest by the reporting researcher). Whichever you use, ask for per-finding records with date, build tested, request, response and status.
Which testing services provide ongoing assessment and re-testing after fixes for a fast-moving startup?
Teams that deploy daily are usually best served by an on-demand option: a pentest-as-a-service platform or an autonomous AI penetration testing tool that can re-run a finding as soon as the fix is deployed. A consultancy retest that depends on a tester's calendar suits a yearly audit cycle better than a weekly release cycle. Many startups combine both.
Is a retest included in the price of a penetration test?
It depends on the provider. Many firms include one retest round within a set period after report delivery, and others charge for additional days. Ask before signing how many rounds are included, how long the retest window is, what happens to findings fixed after it closes, and whether the output is an updated report or a separate letter.
Can we verify fixes ourselves instead of paying for a retest?
You should verify internally first, ideally with a regression test that reproduces the exploit. For audit or customer evidence, though, self-verification carries little weight, because the person who wrote the fix tends to test only the case they fixed. Independent verification by the original tester or an equivalent tool is what makes the result credible.
Ready to Secure Your Application?
Continuous, AI-powered penetration testing for web apps and APIs.
Run your first scan →