Skip to main content

Scanners

Most security programs are a pile of tools with a pile of configs, and nobody can say which version ran last Tuesday. In Dwaar, there's one page: per resource type and per category, one tool and one version.

Open Scan Center, then Scanners. Every scan runs all the categories switched on here, on one snapshot of a resource.

Prerequisites​

  • scanners:read to see this page.
  • scanners:update to change anything.

What runs​

Code​

CategoryToolWhat it checks
SASTOpengrepSecurity bugs in your source, without running it
SecretsGitleaksPasswords, keys and tokens in the repository, including old commits
Dependencies (SCA)TrivyKnown vulnerabilities in the libraries you depend on
Malicious packagesOSV-ScannerDependencies known to be malicious
AI code & MCP configOpengrepUnsafe tool use and injection risk in agent code and MCP configuration
Skills & agent instructionsCisco skill-scannerPrompt injection in skills and files such as AGENTS.md
Model filesModelAuditModel files that run code when loaded
CI workflow securityzizmorWays an outsider could run code or steal build secrets
MisconfigurationsCheckovInsecure settings in infrastructure as code
LicensesScanCodeLicenses of your code and dependencies
Provenance & signaturesprovenance-scannerWhether npm and PyPI packages came from their claimed source
Crypto assetsCBOMkit-theiaCertificates, keys and TLS settings
Crypto usageOpengrepThe algorithms your code calls
Code qualitySonarQubeBugs, code smells, duplication and complexity
Privacy & PIIpii-scannerThe personal data your code handles
Repo settings postureLegitifySecurity settings of your GitHub organization and repositories
Product contextDwaar ContextHow the code is built and where it ships, for lineage and product suggestions

Artifact​

CategoryToolWhat it checks
OS & package vulnerabilitiesTrivyKnown vulnerabilities in the packages inside an image
SBOMTrivyThe full list of components
SecretsKingfisherTokens and keys in image layers, including files a later layer deleted
Crypto assetsCBOMkit-theiaCertificates, keys and TLS settings in the image
AttestationscosignWho built an artifact, and whether it was altered
Model file safetyModelAuditUnsafe model formats
Base image freshnessbase-image-scannerThe base image, even without a Dockerfile, and whether it's stale
Privacy & PIIpii-scannerPersonal data in the application files
Product contextDwaar ContextWhere the image came from and what it runs, for lineage and product suggestions

Cloud​

CategoryToolWhat it checks
Cloud posture (AWS)ProwlerThe security configuration of an AWS account in the regions you chose
Cloud posture (Vercel)ProwlerThe security settings of a Vercel team and its projects
Cloud posture (Railway)Dwaar Railway checksThe security settings of a Railway workspace, its projects and services
Cloud posture (Cloudflare)ProwlerThe security settings of a Cloudflare account's zones and their DNS records

The other resource types have no scanner categories yet.

note

The tools and versions come from Dwaar's catalogue and change over time. This page always shows what's current for your tenant.

Categories that follow another​

A few rows show Follows with a lock. They have no scanner of their own: they run as part of another category, with its tool, version and switch.

CategoryFollows
Crypto libraries (Code)Dependencies (SCA)
Licenses (Artifact)SBOM
Crypto libraries (Artifact)SBOM

If you want to change one of these, change the category it follows.

Change a category​

  1. Pick a resource type with the tabs at the top.
  2. Turn on Descriptions if you want each category explained in place.
  3. Use the switch on a row to turn that category on or off.
  4. If a row offers other tools, choose one to replace the current one. There's one scanner per category, so switching replaces it.
  5. If a row offers more than one version, choose the version to pin.

Most categories run one tool at one version today, so you'll usually see the version as plain text and not as a choice.

Your choices apply to new runs. A run keeps the choices it was queued with, so a change doesn't alter a run already waiting, and results stay explainable after the fact.

Important

Turning a category off means fewer findings, not fewer problems. If you're switching off a category to quiet the list, ignoring specific checks is almost always the better tool.

Product context​

This category finds nothing wrong. It collects facts about how a resource is built and deployed, so Dwaar AI can suggest lineage and products from evidence and not just from names.

The Product context row on the Code tab of Scanners, switched on, with its description.

OnIt readsIt keeps
CodeDockerfiles, CI/CD workflows and Makefiles, Compose, Kubernetes and Helm files, Terraform, package manifests, CODEOWNERS, the README titleBase images and exposed ports, images a pipeline pushes and where it deploys, workload and service names, package names, owning teams
ArtifactThe image's configuration, from Trivy's reportSource and title labels, repository, exposed ports, the names of environment variables, what it runs

It never keeps file contents, script bodies, environment variable values, command arguments or build history, so a secret in a CI file has no way in. No AI model is involved in collecting these facts. Extraction is plain parsing, and the only model that sees them is the one you ask for suggestions.

Three things behave differently from other categories:

  • It takes effect at once when you turn it off. Facts already collected stop going to Dwaar AI straight away, and new scans stop collecting them. Facts already collected are not deleted.
  • It doesn't make anything count as not fully scanned. Switching it on queues no rescans. A resource gets its facts on its next full scan, such as a new commit or a Force rescan.
  • On Artifact, Trivy keeps running while it's on, because that's where the facts come from. If you turned every other Artifact category off to stop image scans, turn this one off too.

When scans run​

This page decides what runs. When it runs is set on each connection's Scanning tab. See How discovery and scans work.

Request a scanner​

Missing a tool or a whole category?

  1. Select Request a scanner.
  2. Pick the resource type and a category, or a new one.
  3. Say which scanner you want and what it should check. A note is required for a new category.
  4. Select Send request.

You should see the category marked Requested, with an ID.

Try this next​

Turn on Descriptions and read the Code tab end to end. Then open a recent run on one of your repositories and see which analyzers actually ran. Comparing the two tells you exactly what your coverage is, in place of what you assumed.