home / products

product

ExploitFinder: AI-assisted penetration testing platform

An AI-assisted analysis system that runs tests following a fixed method. It stops whenever a human decision is needed.

the six stages

Each stage has one job and does not do the others' jobs.

A1 · scope

Scope

Reads the request and the signed scope. Marks what is inside and what is outside.

A2 · recon

Recon

Maps the verified domain: pages, endpoints, visible roles.

A3 · plan

Plan

Proposes test hypotheses, each with a written reason.

A4 · test

Test executor

Runs only actions from a closed list. Logs every request and response.

A5 · critic

Critical review

Tries to break conclusions before they reach a person.

A6 · report

Report writer

Drafts the report with the steps to reproduce each vulnerability.

human control

Three points where a person decides.

01

Before testing

Scope and hypotheses are approved by a person. A hypothesis outside the scope is not run.

02

After testing

A vulnerability is confirmed only if the finding is reproducible and a person has assigned the severity.

03

Before the report

No document reaches the client without review. The reviewer can change the severity or discard the point.

the platform

It starts from the authorised scope.

Every test begins by naming the target, which must fall within an authorised scope. You can set a priority (IDOR/BOLA, SSRF, RCE), add instructions and, if needed, provide authentication.

ExploitFinder screenshot: the What do you want to test form with the target field, the IDOR/BOLA, SSRF and RCE priorities, additional instructions and optional authentication
Starting a test. The target must be in an authorised scope. Priorities and instructions guide the engine, but do not replace the signed scope.

Screenshot of the ExploitFinder platform.

product status

What is ready and what is not.

This is the architecture planned for the services. What is still a prototype or on the roadmap is stated as such.

PartStatus
Action log with chain of custodyprototype
Scope control and domain verificationprototype
Human review and verdict for each vulnerabilityprototype
Analysis connected to a language modelin progress
Public benchmark on test applicationsin progress
Emergency stop for the whole scopeto complete before the first service
Client portal to follow testsroadmap

Want to see the flow on one of your applications?

It starts from a written proposal, with the scope signed before any test.

Request access