Explainer · Cybersecurity testing

All testing possible,
for every target provided.

SCA, SAST, DAST, fuzzing and penetration testing run together, so every result arrives with the context to prove whether it is exploitable.

DISCRETE TESTS VERSUS ONE TESTRun separately, five tools produce five lists with no shared context. Run together, they produce one set of findings with a verdict on each.DISCRETE TESTS · SILOSSCA1,900 CVEsSAST640 findingsDAST88 findingsFuzzing31 crashesPentest14 findingsFive reports and no shared model of the product,so no one can say which of the 2,673 results reach anything.Exploitability analysis cannot be performed comprehensively.ELTON · ALL TESTS, ONE TARGET, ONE MODELSCASASTDASTFuzzingPentestDigital twin + exploitability analysisevery result placed on the product's components, interfaces and dataflows, then verified25 exploitabledirect or conditional · fix these2,648 weaknessesno path on this device · documentedOne report. Every finding carries the test case that found it and the verdict that rates it.DISCRETE TESTS VERSUS ONE TESTFive separate lists, or one set of findingswith a verdict on each.DISCRETE TESTS · SILOSSCA1,900CVEsSAST640findingsDAST88findingsFuzzing31crashesPentest14findingsFive reports, no shared model of the product:nobody can say which of 2,673 results reachanything.ELTON · ALL TESTS, ONE MODELSCASASTDASTFuzzingPentestDigital twin + exploitability analysisevery result placed on the product, then verified25 exploitablefix these2,648 weaknessesdocumented, no pathOne report. Every finding carries the test casethat found it and the verdict that rates it.
The industry has moved away from discrete tests: separate SCA, SAST and DAST engagements leave every result without the context to rate it.
What runs

Every test type,
on every target.

The target decides the scope, not the statement of work.

WHAT RUNS ON EVERY TARGETEvery test type the target supports is executed. The target decides the scope, not the statement of work.FirmwareEmbeddedsoftwareWeb / cloud/ APIMobileNetwork /protocolsHardwareinterfacesSCA / SBOMknown CVEs in every componentSASTsource and compiled binariesDASTrunning services and interfacesFuzzingmalformed input on every parserAttack surfaceentry points, trust boundariesRobustnessload, fault and failure behaviorPenetration testhuman techniques, encodedExploitability verificationtwin · code · real hardwareFilled: executed whenever the target has that surface. Open: not applicable to that target class. Test cases are generated per release from the digital twin.WHAT RUNS ON EVERY TARGETEvery test type the target supports isexecuted.SCA / SBOMknown CVEs in every componentFirmwareEmbeddedWeb/cloudMobileNetworkHardwareSASTsource and compiled binariesFirmwareEmbeddedWeb/cloudMobileNetworkHardwareDASTrunning services and interfacesFirmwareEmbeddedWeb/cloudMobileNetworkHardwareFuzzingmalformed input on every parserFirmwareEmbeddedWeb/cloudMobileNetworkHardwareAttack surfaceentry points, trust boundariesFirmwareEmbeddedWeb/cloudMobileNetworkHardwareRobustnessload, fault and failure behaviorFirmwareEmbeddedWeb/cloudMobileNetworkHardwarePenetration testhuman techniques, encodedFirmwareEmbeddedWeb/cloudMobileNetworkHardwareExploitability verificationtwin · code · real hardwareFirmwareEmbeddedWeb/cloudMobileNetworkHardwareGreen: executed when the target has that surface.

Hardware, firmware, software

Embedded targets get SAST on source and compiled binaries, firmware and bootloader analysis, and fuzzing on every interface, including JTAG, UART and BLE.

Web, cloud, mobile

The services and apps around the device get DAST on running services, API and session testing, and the same SCA and SAST as the device.

Protocols

DICOM, HL7, MQTT and proprietary protocols are fuzzed with malformed traffic on the real product, in our lab or over TestLink.

Why together

Discrete tests
create silos.

Each test answers a different question; only all of them answer the reviewer's.

WHAT EACH TEST CONTRIBUTESNo single test answers the question a reviewer asks: is this exploitable on this device? Together they do.SCA / SBOMwhat is in itSASTwhere the code is weakDASTwhat is reachableFuzzingwhat breaks under bad inputPentestwhat an attacker can chainVerificationwhether it works on the deviceall six results land on the same digital twin, per releaseEXPLOITABLEdirect or conditional, with the test case that proves itWEAKNESSno path on this device, with the rationale for the recordWHAT EACH TEST CONTRIBUTESNo single test answers whether a result isexploitable on this device. Together theydo.SCA / SBOMwhat is in itSASTwhere the code is weakDASTwhat is reachableFuzzingwhat breaks under bad inputPentestwhat an attacker can chainVerificationwhether it works on the deviceall six land on the same digital twin, per releaseEXPLOITABLEdirect or conditional, with the test case that proves itWEAKNESSno path on this device, with the rationale recorded

SCA alone is a version match

A CVE in a component says nothing about whether the vulnerable function is reachable on this device.

SAST alone is a code path

A weak pattern in source may sit behind a control, or in code that never executes on the shipped configuration.

DAST or a pentest alone is a symptom

A reachable fault with no source context cannot be traced to its root, so it cannot be fixed once or chained correctly.

The outcome

Exploitable findings,
documented weaknesses.

Every result is placed on the digital twin and verified before it is called a finding.

FROM RESULTS TO FINDINGSOnly a verified exploit path turns a result into a finding. Everything else is recorded as a weakness with the reason attached.2,673 raw resultsSCA, SAST, DAST, fuzzing, pentest, one target1,410 after deduplicationthe same issue found by three tools is one finding212 reachable on this releaseplaced on the twin: an entry vector leads to it25 exploitable6 direct, 19 conditional, verified on the device2,648 weaknesses stay in the record with a written rationale, because FDA reviewers probe the findings you did not fix.FROM RESULTS TO FINDINGSOnly a verified exploit path turns a resultinto a finding.2,673 raw resultsSCA, SAST, DAST, fuzzing, pentest,one target1,410 after deduplicationthe same issue found by threetools is one finding212 reachable on this releaseplaced on the twin: an entryvector leads to it25 exploitable6 direct, 19 conditional, verifiedon the device2,648 weaknesses stay in the record with a writtenrationale, because reviewers probe what you didnot fix.
How a test runs

Six phases,
one product model.

The same phases every time; open-box access is what lets phase three cover the whole product.

1

Test plan

Scope, constraints, threat focus, test cases generated from the twin.

2

Attack surface

Interfaces, entry points, trust boundaries, workflows.

3

Identify weaknesses

SCA, SAST, DAST, fuzzing and pentest techniques, all at once.

4

Exploit single issues

Verify each candidate on the twin, in code, on the device.

5

Chain and bypass

Multi-step exploitation and root-cause chaining.

6

Reporting

Findings, weaknesses, evidence, prescriptive fixes.

Open-box by default

Source, build, architecture, threat model, SBOM and privileged test access; nearly all premarket tests are now open-box.

Traceable to the twin

Every test case maps to a component, interface or dataflow, which is what FDA reviewers ask to see.

Retested every release

The record is retained, so a new release tests the change against everything already known.

In the report

The methodology section
shows the work.

Every Product Security Testing Report documents the process a reviewer will ask about.

Test types per target

Which of SCA, SAST, DAST, fuzzing, attack surface, robustness and pentest ran against each component, and with what coverage.

Test cases and traceability

Each test case mapped to the architecture element and threat it exercises, with the result recorded.

Verdict and evidence

For every finding the exploit path and proof; for every weakness the rationale for the record.

Questions

Questions about
cybersecurity testing.

Which tests are included in an ELTON test?

Every test type the target supports: software composition analysis against the SBOM, static analysis of source and binaries, dynamic analysis of running services, fuzzing of every parser and protocol, attack surface and robustness analysis, penetration testing, and exploitability verification on the twin, in code and on real hardware.

Why not buy SAST, DAST and a pentest separately?

Because each one answers a different question and none of them answers the reviewer's question. SCA says what is in the product, SAST where the code is weak, DAST what is reachable, fuzzing what breaks, pentesting what an attacker can chain. Run in silos they produce five lists with no shared context, so exploitability cannot be analyzed comprehensively.

What is the difference between a finding and a weakness?

A finding is exploitable on this device as shipped, directly or conditionally, and carries the test case that proves it. A weakness is a real result with no exploit path on this release; it stays in the record with a written rationale, because FDA reviewers probe the results you did not fix.

Where is this documented?

In the methodology section of every Product Security Testing Report: the test types executed per target, the tools and coverage record, test cases mapped to the architecture and threat model, and the verdict and evidence behind every finding and weakness.

Exploitability management for medical devices. FDA §524B methodologyExploitability proven at runtime95% faster than legacy testing Book a Demo
Platform
OverviewAvoid FDA DeficienciesAvoid Consulting FeesDigital Twin TraceabilityAI MedDevice PentestingExploitability VerificationVulnerability ChainingRemediation OptimizationRemote TestLink™Incident ResponseAutomated VEX & MetricsCVSSv4 Migration
Solutions
EnterpriseStartups / SMBs Postmarket SurveillanceIncident ResponseSecurity EngineeringRegulatory AffairsFDA §524BEU MDR/CRAEU REDNIS2IMDRF N60 / N73Japan MHLW
Why ELTON
One Solution Subscription TestingAI-NativeFDA ComplianceVerified ExploitabilityELTON vs. Legacy TestingThreat-Led AI PentestingMDDT MethodologyCredentialsDevice ModalitiesPricing
Resources
FDA Deficiency ListFDA Testing RequirementsFDA Cyber SOPs & TemplatesRemediation LibraryRegulatory GuidesWebinarsAI NewsletterThe End of Legacy TestingThe AI Vulnerability ExplosionAI Inside the ProductCybersecurity TestingSecurity AdvisoriesWhitepapersIntelligence & Blog
Company
AboutLeadershipCareersPartnershipsData SecurityContact Meet ELTON
See the graph decide, live >