What only shows when it runs
Misconfigurations, injectable parameters, broken authentication and missing security headers show up in the deployed application, not in a single source file.
VamiAppSec runs VamiDAST, our AI-powered DAST, as a fixed stage in your pipeline, on demand via CLI or in production in Passive or Safe mode. It tests the running web app and API across 30+ vulnerability classes of the OWASP Top 10 and the OWASP API Security Top 10 — and puts every confirmed finding, with payload and evidence, into the same backlog as your static results.
Some vulnerabilities only exist at runtime — in configuration, authentication flows or the interplay of services. Four reasons to test the running build on every release:
Misconfigurations, injectable parameters, broken authentication and missing security headers show up in the deployed application, not in a single source file.
Broken object-level authorization (BOLA/IDOR) needs two identities and real requests — VamiDAST tests with a second user to prove it.
Regulations and standards expect regular testing — for example CRA Annex I, Part II (3), IEC 62443-4-1 SVV-3 or ISO/SAE 21434 clause 10.
A pentest is a snapshot. A DAST stage scans every build that reaches staging, so new exposures are caught before release.
A deterministic engine does the testing; the LLM advises in three bounded places.
| Step | What VamiDAST does | Result |
|---|---|---|
| Crawl attack surface | katana and a headless browser map the web app and API from HTML, JavaScript, OpenAPI specs, robots.txt and wordlists — including authenticated flows. | Endpoints with parameters and JSON bodies |
| Classify heuristics first | Path and parameter heuristics assign each endpoint its likely vulnerability class; the LLM is consulted only as a last resort. | Targeted tests instead of blind fuzzing |
| Attack safety interlock | Deterministic detectors and an injection sweep, plus sqlmap, dalfox, commix and Nuclei where a tool fits — a safety interlock gates every active request. | Payloads matched to each endpoint |
| Confirm second pass | Unambiguous signatures decide first; the LLM triages only the grey zone, and a second pass re-probes anomalies. | Reported as confirmed or as anomaly |
| Track and gate DefectDojo | Findings arrive normalized — payload, evidence, confidence, remediation — next to SAST and SCA results. | Confirmed findings can fail the test gate |
| Prove OWASP WSTG · ASVS | Findings carry OWASP WSTG test IDs, feed the ASVS coverage and join the per-release evidence bundle. | Traceable evidence for auditors |
The same engine, the same safety modes and the same backlog — wherever the scan starts.
A fixed stage scans every build that reaches staging; confirmed findings at or above your severity threshold can fail the test gate.
Start a scan from the terminal before a release, after a hotfix or for a one-off assessment of any authorized target.
Continuous or scheduled scans of live applications in Passive or Safe mode; new exposures land in the same backlog.
Forms, JWT, cookies and API keys, plus a second identity for BOLA/IDOR — credentials are never stored, logged or put into a prompt.
Coverage telemetry keeps a silent zero from passing as a clean result.
On T Cloud Public, exclusively in German data centers operated by Deutsche Telekom.
VamiDAST, the other scanners, DefectDojo and the LLM on your own Kubernetes cluster — close to the applications you test.
Crawling, tool runs and detection are deterministic. The LLM acts in three bounded places, every output is schema-validated and only 2 KB response snippets reach a prompt.
VamiDAST is also available as a product of its own: vamidast.com. In VamiAppSec it is fully integrated — same backlog, same gates, same evidence bundle.
Dynamic application security testing examines a running web application or API from the outside, the way an attacker would: it sends requests and analyses the responses. Unlike SAST it does not need the source code, and it finds runtime issues such as misconfigurations or broken authentication.
Yes, in Passive or Safe mode. Passive only crawls and inspects; Safe tests read-only endpoints fully and state-changing ones only via precondition violation. Aggressive mode is only available after explicit confirmation that the target is not production.
More than 30 vulnerability classes across the OWASP Top 10 and the OWASP API Security Top 10 — such as injection, cross-site scripting and broken object-level authorization — plus passive posture checks in every mode.
A limited one, by design: the engine is deterministic, and the LLM acts only in three bounded places — classifying an endpoint, drafting fallback payloads and triaging the grey zone. Every output is schema-validated.
ZAP by Checkmarx can run as an additional DAST engine in VamiAppSec. VamiDAST is the integrated, AI-powered engine; both feed the same backlog.
Yes. Automated DAST covers known vulnerability classes on every build; a manual penetration test adds creativity, business-logic abuse and chained attacks. VamiSec, the company behind VamiAppSec, offers penetration testing as a service.
In 30 minutes we run the pipeline from SAST to VamiDAST against a staging URL and show you confirmed findings with payload and evidence, the test gate and the DefectDojo board.