Your tests stay Playwright, Selenium and pytest. We don't replace them. We make them enterprise-ready, and you own your code, your evidence and your history.

Smart Monkey testing in Nexus StudioQA.

Smart Monkey testing in Nexus StudioQA is an autonomous security scan of a web application you are authorized to test. It crawls the target, submits forms, sends non-destructive payloads and records each finding with a severity, a screenshot and a replay script. It runs on your own machine, exits 1 when it finds something, and writes SARIF for CI.

By the Nexus StudioQA team, Verdict System LLC.

Last updated: .

What it looks for

Every finding type has a fixed severity inside the scanner. The SARIF export reads CRITICAL and HIGH as error, MEDIUM as warning and the rest as note.

SeveritySARIF levelFinding types
CRITICALerrorXSS execution, SQL injection (including blind, boolean-based and time-based), command injection, server-side template injection
HIGHerrorBroken access control, 5xx server errors, CORS misconfiguration, exposed sensitive files, SSRF, confirmed IDOR, path traversal, XXE, mass assignment, insecure deserialization, blind XSS confirmed by callback
MEDIUMwarningClient-side runtime errors, insecure cookie flags, missing security headers, open redirects, GraphQL introspection left on
LOWnoteAn object reference that looks like an IDOR but was seen from one session only, and a scan that failed to run

How a scan runs

The scan is a list of phases. By default it runs all of them headless and stops at the first bug it records. The agent phase is opt-in and is not part of the default run.

CommandWhat it does
nexus monkey <url>Scans every phase and stops at the first bug (FAST mode).
--deepRecords a bug, isolates it and keeps searching for more.
--phase=<name>Runs one phase: pre-auth, post-forms, auth-transit, post-auth, http-probes, ssrf, cmdi, xss-oob, sqli, ssti, lfi, xxe, graphql, deser, mass-assign, idor or agent.
--seed=<n>Replays the random choices of an earlier run. Every run prints its seed.
--mobileEmulates an iPhone: user agent, a 390 by 844 viewport and touch.
--chaos[=ms]Adds latency to every request, 3500 ms when no value is given.
--headedShows the browser instead of running headless.

What keeps it contained

  • It asks for authorization first

    A named operator types AUTHORIZED for a named target before a scan starts, and the record is written into your own evidence tree. A refused scan exits 2 and scans nothing.

  • Payloads are non-destructive

    A filter drops destructive SQL such as DROP, DELETE, UPDATE and INSERT before anything is sent, the scanner skips forms and URLs whose wording suggests a state-changing action, and the HTTP-layer probes use GET only. It still submits forms, so point it at a test environment.

  • Blind findings are confirmed on your side

    SSRF, blind XSS and command injection leave no trace in the response. The scanner runs its own listener, puts a unique URL in the payload and confirms the finding when the target calls back. No public callback service is involved, so no URL or payload goes to a third party.

  • An incomplete scan is never clean

    Exit 0 means clean, exit 1 means a vulnerability was recorded, exit 2 means the scan did not complete or was not authorized.

What it leaves behind

One session folder under evidence/monkey-audits/, with a folder per finding. In the Studio, Smart Monkey is on the Security screen under Run a scan. The newest 10 sessions are kept, and NEXUS_MONKEY_RETAIN changes that.

FileWhat it is
audit_report.jsonThe finding as a machine-readable record.
screenshot.pngThe page at the moment of the finding.
replay.webmThe session recording, copied into each finding.
replay.mjsAn executable reproduction of the finding.
session_summary.jsonOne index of the whole session.
results.sarifSARIF 2.1.0, one result per finding, for GitHub code scanning or GitLab. nexus monkey writes it on a CI runner without the Studio.

What a clean run does not mean

Exit 0 means this set of probes found nothing on the pages the scan reached. It is a scanner, not a penetration test, and it sees only what it can crawl with the session it has. The API audit and the API authentication replay are opt-in (NEXUS_API_AUDIT, NEXUS_API_ACTIVE), and so is the AI triage of findings (NEXUS_SECURITY_AI).

Common questions

  • Can I run Smart Monkey against any website?

    No. Only against systems you own or are authorized to test. The scan will not start until AUTHORIZED is given for the target. Acceptable use

  • Does it change or delete data on the target?

    The payload filter excludes destructive SQL and the scanner skips state-changing forms, but it does submit forms and send test payloads. Run it against staging or another test environment, not production data.

  • Can it run in CI?

    Yes. nexus monkey runs without the Studio, writes results.sarif and returns exit code 0, 1 or 2, so a pipeline can gate on it.

  • Does scan data leave my machine?

    The scan sends traffic only to the target you name. Blind findings are confirmed by a listener on the scanning machine, and findings are stored in your evidence folder. The only connection the product makes to us is the license check.

Run it on your own machine.

Try every feature free, on your own machines. No card needed.