Penetration Testing Remediation: Handling Critical Findings
Handle penetration test findings in a fixed order: contain anything exploitable right now, confirm each finding and its real impact, assign an owner and a deadline by severity, fix the root cause instead of the reported payload, then have the fix independently retested before you close the ticket. Record any accepted risk with a named owner and a review date.
Continuous AI-powered penetration testing by Penetrify - find and fix vulnerabilities before attackers do.
Step 1: Contain what is exploitable today
A critical finding in a penetration test report means someone demonstrated the attack. Assume others can too. Before anyone designs a proper fix, reduce the exposure:
- Rotate anything the report exposed. API keys, database passwords, tokens or session secrets that appear in the evidence are compromised, including by the report itself as it gets forwarded around.
- Apply a temporary block. Options include a WAF rule for the specific request pattern, disabling the affected feature behind a flag, restricting an admin interface to internal addresses, or taking an unused endpoint offline.
- Check whether it was already used. Search access and application logs for the request pattern from the proof of concept, outside the tester's addresses and test dates. If you find hits, this stops being remediation and becomes incident response.
- Limit who has the report. It is a working exploit manual until the fixes ship.
Containment is not the fix. Record each temporary measure in the ticket so it is removed or made permanent on purpose.
Step 2: Confirm each finding and its real severity
Have an engineer who knows the system reproduce every critical and high finding from the report's steps. This does three things: it proves the finding is real, it shows the developer exactly what goes wrong, and it gives you the "before" state for the retest.
Then check the rating against your context. The tester scored the issue with limited knowledge of your business. Ask:
- What data or action does this actually reach? Customer records and payment flows rank above an internal status page.
- Does it need authentication, a specific role or user interaction?
- Is the affected system reachable from the internet?
- Does it chain with another finding? Two mediums that combine into account takeover are a critical.
Adjust severity up or down where the answers justify it, and write the reason down. If you think a finding is a false positive, send the tester your evidence and let them withdraw or defend it. Do not close it quietly.
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 →Step 3: Give every finding an owner and a deadline
Create one ticket per finding in the tracker your engineers already use. A PDF in a shared drive is not a remediation plan. Each ticket needs a single named owner (a person, not a team) and a due date set by severity.
Pick deadlines you will keep. A common shape is days for criticals, a few weeks for highs, the next quarter for mediums and best effort for lows. If a customer contract or compliance framework sets a deadline, that one wins. A ticket that carries everything a developer needs looks like this:
Title: [PT-07] IDOR on GET /api/invoices/{id}
Severity: Critical (tester: High; raised, exposes all tenants)
Owner: one named engineer
Due: 2 business days
Affected: invoices-service, all /api/invoices/* routes
Repro: report p.14, steps 1-4 (confirmed on staging)
Mitigation: WAF rule 1182 active since day one, remove after fix
Fix plan: tenant ownership check in shared middleware
Verify: regression test + retest by original tester
Step 4: Fix the root cause, not the reported payload
The report shows one example of each weakness. The tester had limited time, so there are usually more instances than the report lists. The fix has to remove the class of bug.
- Injection (SQL, command, template). Use parameterized queries or safe APIs everywhere the pattern occurs. Filtering the string from the proof of concept leaves the flaw in place.
- Broken access control and IDOR. Enforce ownership and role checks on the server, in one shared layer, for every route that takes an object ID. Then test the sibling endpoints yourself.
- Cross-site scripting. Encode output for its context by default in the templating layer, and add a content security policy as a second line.
- Authentication and session flaws. Fix the mechanism (token validation, session invalidation on logout or password change, rate limiting) and then expire existing sessions.
- Exposed secrets and misconfiguration. Rotate first, then change the process that put the secret there.
- Vulnerable components. Upgrade, and confirm the vulnerable code path is no longer reachable.
For every fix, add an automated test that reproduces the original exploit and fails on the old code. Search the codebase for the same pattern and fix those instances in the same change or in linked tickets.
Step 5: Decide what to do with findings you cannot fix yet
Some fixes need a redesign, a vendor patch or a migration that will not land this month. There are two legitimate options, and "leave it open" is not one of them.
- Compensating control. Reduce the likelihood or impact another way: network restriction, extra monitoring and alerting on the attack pattern, a feature flag, a stricter rate limit. Document what it covers and what it does not.
- Formal risk acceptance. A named person with authority over the affected system signs off, with the reason, the compensating controls in place and a review date. An acceptance without an expiry date turns into permanent debt.
Low and informational findings can often be accepted or batched into normal maintenance. Criticals should almost never be accepted. If one is, the decision belongs with executive leadership, not with the team that owns the backlog.
Step 6: Retest, close and report out
A finding is closed when an independent retest shows the attack fails, not when the pull request merges. Ask your testing provider how retests work before you start fixing: how many rounds are included and how long the window stays open. Send the tester a short note per finding saying what changed, and keep the test accounts and data they used.
If the original test came from an automated platform, verification can be immediate. Penetrify, for example, includes free targeted re-tests of fixed findings, so each fix can be checked as it deploys. When a contract requires a retest letter from a named third-party firm, that firm still has to perform it.
Once retesting is done, produce a one-page remediation summary: each finding, its status (fixed, risk accepted, in progress with date) and the evidence reference. This is what customers and auditors ask for, whether the test was for SOC 2, PCI DSS or a security questionnaire. Share that summary instead of the full report where you can.
Step 7: Use the findings to prevent the next report
Before the report is archived, spend an hour on why each critical existed. The answers usually fall into a few groups: no shared authorization layer, no review step for a risky kind of change, a framework default nobody questioned, a secret handled by hand.
Turn each into one concrete change: a library, a lint rule, a pipeline check or a line in the code review checklist. Then shorten the gap until the next test. If the application changes weekly, an annual test finds a year of accumulated issues at once, which is how reports full of criticals happen. Automated testing between manual engagements keeps the batch small. Penetrify runs autonomous tests against web applications and APIs from a CI/CD pipeline. A manual test remains the better fit for deep business logic review and for anything that needs a human tester's signed report.
Frequently Asked Questions
How do I handle penetration test findings, and what is the remediation process?
Work in order: contain critical exposure with temporary measures, reproduce and re-rate each finding, create one ticket per finding with a named owner and a deadline, fix the underlying class of bug, document compensating controls or risk acceptance for anything you cannot fix, and get an independent retest. Finish with a short remediation summary you can show to customers and auditors.
How quickly do critical penetration test findings need to be fixed?
No universal deadline exists. Many teams aim to fix or mitigate criticals within days and highs within a few weeks, and a customer contract or compliance framework may set its own limit. What matters most is that containment happens immediately, the deadline is written on the ticket, and a missed deadline triggers escalation instead of silence.
What should we do first when a pentest finds a critical vulnerability?
Reduce exposure before designing the fix. Rotate any credentials shown in the report, apply a temporary block such as a WAF rule or a feature flag, and search your logs for the same attack pattern outside the test dates. If the logs show earlier use by someone else, treat it as a security incident, not as routine remediation.
Do we have to fix every finding in a penetration test report?
No, but every finding needs a recorded decision. Critical and high findings should be fixed or covered by a compensating control. Lower severity findings can be scheduled into normal maintenance or formally accepted. An acceptance needs a named approver, a reason and a review date, so that an auditor can see it was a decision and not an oversight.
What if we think a penetration test finding is a false positive?
Try to reproduce it using the steps in the report. If it does not reproduce, or the impact is not what the report claims, send your evidence to the tester and ask them to withdraw, re-rate or defend the finding. Keep the exchange with the ticket. Closing a disputed finding without the tester's agreement leaves a gap in the audit trail.
Ready to Secure Your Application?
Continuous, AI-powered penetration testing for web apps and APIs.
Run your first scan →