v2.0
LLM support across static analysis, the Credentials tab, and credential bruteforcing.
The release that first brought language models into the scanning path, and gave discovered credentials somewhere to live.
LLM support in static analysis
Both cloud-hosted and locally hosted models became usable across several static analysis modules:
| Module | What the model added |
|---|---|
| SAST | Model-assisted analysis alongside the pattern-based scanners. |
| Secrets verification | A model judging whether a detected secret is a real leaked credential rather than an example or a long-rotated test key. |
| Dependency check enrichment | Advisory findings enriched with context beyond the raw CVE record. |
Local models were supported from the start of this work. Any endpoint implementing the OpenAI API specification is usable, so a sensitive codebase never has to leave your infrastructure.
Custom rule generation
Nuclei rules could now be generated from a vulnerability proof-of-concept description rather than written by hand. Paste the description — including the relevant parts of the HTTP request and response — and the model drafts a rule you review, amend and upload. Covered in Custom Rules.
New: the Credentials tab
A dedicated section for credentials the platform finds or you supply:
- Leaked credentials identified during OSINT scans.
- Credentials found through external data-leak monitoring services.
- Credentials discovered by code secrets scans, saved automatically rather than left in a report.
- Bulk upload in
.jsonand.xlsxformat.
See Credentials.
Bruteforce module
The bruteforcer enumerates known network services — FTP, SSH, SMB and others — and attempts authentication using:
- Credentials saved in the Credentials tab, to verify whether a leaked pair actually still works across the network.
- Built-in wordlists of common and default credentials.
This closes a loop that most platforms leave open: a secret found in code is a suspicion, and a secret confirmed to authenticate against a live service is a finding. Confirmed credentials are marked verified in the Credentials tab.
Additional scanners
New web and infrastructure scanners were added for broader detection coverage.
Last reviewed 2026-08-16