ModernPentestModernPentest
AI Agents

Validation Agents

AI agents that verify findings and confirm remediations

Agent Overview

AgentPurposeWhen Activated
Finding Triage AgentValidates and triages reported findingsAfter testing agents report vulnerabilities
Remediation Verification AgentConfirms vulnerability fixesOn-demand and before each pentest

Finding Triage Agent

Every vulnerability reported by testing agents passes through the Finding Triage Agent before reaching your dashboard. This ensures you only see validated, properly categorized issues.

What It Does

FunctionDescription
False Positive FilteringValidates each finding is a real, exploitable vulnerability
Severity AssessmentAssigns accurate severity based on actual exploitability and impact
Category VerificationEnsures correct CWE and OWASP classification
DeduplicationMerges related findings from multiple agents
Evidence ReviewConfirms proof-of-concept evidence is complete

How It Works

Receive Findings

Testing agents (Access Control, Injection, Auth, Scanner) report potential vulnerabilities to a shared queue.

Validate Exploitability

The Triage Agent examines each finding's proof-of-concept and confirms the vulnerability can be exploited in the application's context.

Assess & Classify

  • Assigns severity using CVSS factors plus actual exploitability
  • Maps to appropriate CWE and OWASP categories
  • Deduplicates related findings from multiple agents

Enrich & Publish

Adds framework-specific remediation guidance. Only validated findings reach your dashboard.

Validation Process

For each finding, the Triage Agent:

  1. Reviews Evidence — Examines the proof-of-concept, request/response pairs, and screenshots
  2. Confirms Exploitability — Verifies the vulnerability can be exploited in the application's context
  3. Assesses Impact — Evaluates real-world consequences (data exposure, privilege escalation, etc.)
  4. Assigns Severity — Uses CVSS factors plus actual exploitability to set accurate severity
  5. Classifies — Maps to appropriate CWE and OWASP categories
  6. Adds Guidance — Includes framework-specific remediation steps

Why This Matters

Traditional scanners often report hundreds of findings with high false positive rates. The Finding Triage Agent ensures:

  • No noise — Only validated, exploitable vulnerabilities reach your dashboard
  • Accurate severity — Findings reflect real-world risk, not theoretical maximums
  • Proper categorization — Each finding correctly maps to industry standards
  • Complete evidence — Every vulnerability includes proof-of-concept for verification

Our goal: less than 5% false positive rate. Every finding you see has been validated by AI before reaching your dashboard.

Remediation Verification Agent

After you fix a vulnerability, how do you know it's actually fixed? The Remediation Verification Agent confirms that your remediation works.

What It Does

FunctionDescription
On-Demand VerificationTests when you mark a vulnerability as remediated
Pre-Pentest VerificationRe-tests all vulnerabilities before each new pentest
Status ManagementTransitions vulnerabilities to Fixed or back to In Remediation
Activity LoggingRecords verification results and evidence

When It Runs

On-Demand (User-Triggered)

When you mark a vulnerability as "Remediated" in the dashboard:

User Triggers Verification

You mark a vulnerability as "Remediated" in the dashboard.

Agent Activates

The Remediation Verification Agent receives the request and loads the original vulnerability details.

Comprehensive Testing

  • Re-runs the original exploit
  • Tests with payload variations
  • Checks for common bypasses

Status Update

Updates the vulnerability status based on results: Fixed if remediated successfully, or back to In Remediation with notes if still vulnerable.

Pre-Pentest (Automatic)

Before each scheduled or on-demand pentest:

Pentest Triggered

A new pentest is scheduled or manually started.

Scan Open Vulnerabilities

The Verification Agent re-tests all open and in-remediation vulnerabilities from previous pentests.

Update Statuses

Vulnerabilities that are now fixed are marked as such. Those still exploitable remain open.

Testing Continues

Testing agents proceed, focusing on finding new vulnerabilities rather than re-reporting fixed ones.

Verification Process

The Remediation Verification Agent thoroughly tests each fix:

  1. Reproduces Original Attack — Runs the exact exploit that found the vulnerability
  2. Tests Variations — Tries bypass techniques and payload modifications
  3. Checks Related Endpoints — Verifies similar patterns aren't vulnerable elsewhere
  4. Documents Results — Records evidence of successful remediation or continued vulnerability

Status Transitions

Original StatusVerification ResultNew Status
RemediatedVulnerability fixedFixed
RemediatedStill exploitableIn Remediation (with notes)

Example Verification

For a SQL Injection vulnerability in /api/search:

TestPrevious ResultCurrent Result
Original payload (' OR '1'='1' --)200 OK with data leak400 Bad Request
URL-encoded payloadData leakBlocked
Unicode bypassData leakBlocked
Double encodingData leakBlocked

Verdict: Fixed with high confidence. All test payloads now properly rejected.

Verification in Activity Timeline

Every verification creates an activity entry on the vulnerability:

  • Agent verdict — Fixed, Still Vulnerable, or Regression
  • Test results — What was tested and outcomes
  • Evidence — Request/response pairs demonstrating current behavior
  • Timestamp — When verification occurred

This provides a complete audit trail for compliance documentation.

Integration with Pentest Workflow

Finding Triage in Testing Pipeline

Discovery Phase

Reconnaissance Agent maps the application, identifying endpoints, technologies, and input points.

Parallel Testing

Multiple testing agents (Access Control, Injection, Authentication, Scanner) test simultaneously, reporting findings to a shared queue.

Triage Phase

Finding Triage Agent validates all findings, filters false positives, assigns severity, and adds remediation guidance.

Report Generation

Only validated findings are included in the final report.

Remediation Verification in Vulnerability Lifecycle

StageDescription
OpenVulnerability found and validated by Triage Agent
In RemediationYou've acknowledged and are working on the fix
RemediatedYou mark as remediated; Verification Agent tests the fix
FixedVerification confirms the vulnerability is resolved

If verification fails, the vulnerability returns to In Remediation with notes explaining what's still exploitable.

SOC 2 Compliance

Both validation agents support SOC 2 requirements:

Finding Triage Agent

Supports Trust Services Criteria CC4.1 (Monitoring Activities):

  • Demonstrates systematic vulnerability assessment
  • Provides evidence of finding validation methodology
  • Documents severity assessment criteria

Remediation Verification Agent

Supports Trust Services Criteria CC7.2 (Remediation of Identified Matters):

  • Proves vulnerabilities were actually fixed
  • Provides evidence of remediation testing
  • Documents the verification methodology
  • Creates audit trail of fix verification

SOC 2 auditors want to see that findings are validated and fixes are verified. Our validation agents provide the documentation they need.

Next Steps

Last updated: January 11, 2026

On this page