Grype logo

Static Analysis + Unit + Integration
Go Report Card
GitHub release
GitHub go.mod Go version
License: Apache-2.0
Slack Invite

A vulnerability scanner for container images and filesystems. Easily install the binary to try it out. Works with Syft, the powerful SBOM (software bill of materials) tool for container images and filesystems.

Join our community meetings!



  • Scan the contents of a container image or filesystem to find known vulnerabilities.
  • Find vulnerabilities for major operating system packages:
    • Alpine
    • Amazon Linux
    • BusyBox
    • CentOS
    • Debian
    • Distroless
    • Oracle Linux
    • Red Hat (RHEL)
    • Ubuntu
  • Find vulnerabilities for language-specific packages:
    • Ruby (Gems)
    • Java (JAR, WAR, EAR, JPI, HPI)
    • JavaScript (NPM, Yarn)
    • Python (Egg, Wheel, Poetry, requirements.txt/ files)
  • Supports Docker and OCI image formats

If you encounter an issue, please let us know using the issue tracker.



curl -sSfL | sh -s -- -b /usr/local/bin

…or, you can specify a release version and destination directory for the installation:

curl -sSfL | sh -s -- -b <DESTINATION_DIR> <RELEASE_VERSION>


brew tap anchore/grype
brew install grype

Note: Currently, Grype is built only for macOS and Linux.

Getting started

Install the binary, and make sure that grype is available in your path. To scan for vulnerabilities in an image:

grype <image>

The above command scans for vulnerabilities that are visible in the container (i.e., the squashed representation of the image). To include software from all image layers in the vulnerability scan, regardless of its presence in the final image, provide --scope all-layers:

grype <image> --scope all-layers

Supported sources

Grype can scan a variety of sources beyond those found in Docker.

# scan a container image archive (from the result of `docker image save ...`, `podman save ...`, or `skopeo copy` commands)
grype path/to/image.tar

# scan a directory
grype dir:path/to/dir

Use Syft SBOMs for even faster vulnerability scanning in Grype:

# Just need to generate the SBOM once
syft <image> -o json > ./image-sbom.json

# Then scan for new vulnerabilities as frequently as needed
grype sbom:./image-sbom.json

# (You can also pipe the SBOM into Grype)
cat ./image-sbom.json | grype

Sources can be explicitly provided with a scheme:

docker:yourrepo/yourimage:tag          use images from the Docker daemon
docker-archive:path/to/yourimage.tar   use a tarball from disk for archives created from "docker save"
oci-archive:path/to/yourimage.tar      use a tarball from disk for OCI archives (from Skopeo or otherwise)
oci-dir:path/to/yourimage              read directly from a path on disk for OCI layout directories (from Skopeo or otherwise)
dir:path/to/yourproject                read directly from a path on disk (any directory)
registry:yourrepo/yourimage:tag        pull image directly from a registry (no container runtime required)

Excluding file paths

Grype can exclude files and paths from being scanned within a source by using glob expressions
with one or more --exclude parameters:

grype <source> --exclude './out/**/*.json' --exclude /etc

Note: in the case of image scanning, since the entire filesystem is scanned it is
possible to use absolute paths like /etc or /usr/**/*.txt whereas directory scans
exclude files relative to the specified directory. For example: scanning /usr/foo with
--exclude ./package.json would exclude /usr/foo/package.json and --exclude '**/package.json'
would exclude all package.json files under /usr/foo. For directory scans,
it is required to begin path expressions with ./, */, or **/, all of which
will be resolved relative to the specified scan directory. Keep in mind, your shell
may attempt to expand wildcards, so put those parameters in single quotes, like:

Output formats

The output format for Grype is configurable as well:

grype <image> -o <format>

Where the formats available are:

  • table: A columnar summary (default).
  • cyclonedx: An XML report conforming to the CycloneDX 1.2 specification.
  • json: Use this to get as much information out of Grype as possible!
  • template: Lets the user specify the output format. See “Using templates” below.

Using templates

Grype lets you define custom output formats, using Go templates. Here’s how it works:

  • Define your format as a Go template, and save this template as a file.

  • Set the output format to “template” (-o template).

  • Specify the path to the template file (-t ./path/to/custom.template).

  • Grype’s template processing uses the same data models as the json output format — so if you’re wondering what data is available as you author a template, you can use the output from grype <image> -o json as a reference.

Example: You could make Grype output data in CSV format by writing a Go template that renders CSV data and then running grype <image> -o template -t ~/path/to/csv.tmpl.

Here’s what the csv.tmpl file might look like:

"Package","Version Installed","Vulnerability ID","Severity"
{{- range .Matches}}
{{- end}}

Which would produce output like:

"Package","Version Installed","Vulnerability ID","Severity"

Gating on severity of vulnerabilities

You can have Grype exit with an error if any vulnerabilities are reported at or above the specified severity level. This comes in handy when using Grype within a script or CI pipeline. To do this, use the --fail-on <severity> CLI flag.

For example, here’s how you could trigger a CI pipeline failure if any vulnerabilities are found in the ubuntu:latest image with a severity of “medium” or higher:

grype ubuntu:latest --fail-on medium

Specifying matches to ignore

If you’re seeing Grype report false positives or any other vulnerability matches that you just don’t want to see, you can tell Grype to ignore matches by specifying one or more “ignore rules” in your Grype configuration file (e.g. ~/.grype.yaml). This causes Grype not to report any vulnerability matches that meet the criteria specified by any of your ignore rules.

Each rule can specify any combination of the following criteria:

  • vulnerability ID (e.g. "CVE-2008-4318")
  • fix state (allowed values: "fixed", "not-fixed", "wont-fix", or "unknown")
  • package name (e.g. "libcurl")
  • package version (e.g. "1.5.1")
  • package type (e.g. "npm"; these values are defined here)
  • package location (e.g. "/usr/local/lib/node_modules/**"; supports glob patterns)

Here’s an example ~/.grype.yaml that demonstrates the expected format for ignore rules:

  # This is the full set of supported rule fields:
  - vulnerability: CVE-2008-4318
    fix-state: unknown
      name: libcurl
      version: 1.5.1
      type: npm
      location: "/usr/local/lib/node_modules/**"

  # We can make rules to match just by vulnerability ID:
  - vulnerability: CVE-2017-41432
  # ...or just by a single package field:
  - package:
      type: gem

Vulnerability matches will be ignored if any rules apply to the match. A rule is considered to apply to a given vulnerability match only if all fields specified in the rule apply to the vulnerability match.

When you run Grype while specifying ignore rules, the following happens to the vulnerability matches that are “ignored”:

  • Ignored matches are completely hidden from Grype’s output, except for when using the json or template output formats; however, in these two formats, the ignored matches are removed from the existing matches array field, and they are placed in a new ignoredMatches array field. Each listed ignored match also has an additional field, appliedIgnoreRules, which is an array of any rules that caused Grype to ignore this vulnerability match.

  • Ignored matches do not factor into Grype’s exit status decision when using --fail-on <severity>. For instance, if a user specifies --fail-on critical, and all of the vulnerability matches found with a “critical” severity have been ignored, Grype will exit zero.

Note: Please continue to report any false positives you see! Even if you can reliably filter out false positives using ignore rules, it’s very helpful to the Grype community if we have as much knowledge about Grype’s false positives as possible. This helps us continuously improve Grype!

Showing only “fixed” vulnerabilities

If you only want Grype to report vulnerabilities that have a confirmed fix, you can use the --only-fixed flag. (This automatically adds ignore rules into Grype’s configuration, such that vulnerabilities that aren’t fixed will be ignored.)

For example, here’s a scan of Alpine 3.10:

apk-tools     2.10.6-r0  2.10.7-r0  CVE-2021-36159  Critical
libcrypto1.1  1.1.1k-r0             CVE-2021-3711   Critical
libcrypto1.1  1.1.1k-r0             CVE-2021-3712   High
libssl1.1     1.1.1k-r0             CVE-2021-3712   High
libssl1.1     1.1.1k-r0             CVE-2021-3711   Critical

…and here’s the same scan, but adding the flag --only-fixed:

apk-tools  2.10.6-r0  2.10.7-r0  CVE-2021-36159  Critical

Grype’s database

When Grype performs a scan for vulnerabilities, it does so using a vulnerability database that’s stored on your local filesystem.

By default, Grype automatically manages this database for you. Grype checks for new updates to the vulnerability database to make sure that every scan uses up-to-date vulnerability information. This behavior is configurable. For more information, see the Managing Grype’s database section.

How database updates work

Grype’s vulnerability database is a SQLite file, named vulnerability.db. Updates to the database are atomic: the entire database is replaced and then treated as “readonly” by Grype.

Grype’s first step in a database update is discovering databases that are available for retrieval. Grype does this by requesting a “listing file” from a public endpoint:

The listing file contains entries for every database that’s available for download.

Here’s an example of an entry in the listing file:

  "built": "2021-10-21T08:13:41Z",
  "version": 3,
  "url": "",
  "checksum": "sha256:8c99fb4e516f10b304f026267c2a73a474e2df878a59bf688cfb0f094bfe7a91"

With this information, Grype can select the correct database (the most recently built database with the current schema version), download the database, and verify the database’s integrity using the listed checksum value.

Managing Grype’s database

Note: During normal usage, there is no need for users to manage Grype’s database! Grype manages its database behind the scenes. However, for users that need more control, Grype provides options to manage the database more explicitly.

Local database cache directory

By default, the database is cached on the local filesystem in the directory $XDG_CACHE_HOME/grype/db/<SCHEMA-VERSION>/. For example, on macOS, the database would be stored in ~/Library/Caches/grype/db/3/. (For more information on XDG paths, refer to the XDG Base Directory Specification.)

You can set the cache directory path using the environment variable GRYPE_DB_CACHE_DIR.

Offline and air-gapped environments

By default, Grype checks for a new database on every run, by making a network call over the Internet. You can tell Grype not to perform this check by setting the environment variable GRYPE_DB_AUTO_UPDATE to false.

As long as you place Grype’s vulnerability.db and metadata.json files in the cache directory for the expected schema version, Grype has no need to access the network. Additionally, you can get a listing of the database archives available for download from the grype db list command in an online environment, download the database archive, transfer it to your offline environment, and use grype db import <db-archive-path> to use the given database in an offline capacity.

If you would like to distribute your own Grype databases internally without needing to use db import manually you can leverage Grype’s DB update mechanism. To do this you can craft your own listing.json file similar to the one found publically (see grype db list -o raw for an example of our public listing.json file) and change the download URL to point to an internal endpoint (e.g. a private S3 bucket, an internal file server, etc). Any internal installation of Grype can receive database updates automatically by configuring the db.update-url (same as the GRYPE_DB_UPDATE_URL environment variable) to point to the hosted listing.json file you’ve crafted.

CLI commands for database management

Grype provides database-specific CLI commands for users that want to control the database from the command line. Here are some of the useful commands provided:

grype db status — report the current status of Grype’s database (such as its location, build date, and checksum)

grype db check — see if updates are available for the database

grype db update — ensure the latest database has been downloaded to the cache directory (Grype performs this operation at the beginning of every scan by default)

grype db list — download the listing file configured at db.update-url and show databases that are available for download

grype db import — provide grype with a database archive to explicitly use (useful for offline DB updates)

Find complete information on Grype’s database commands by running grype db --help.

Shell completion

Grype supplies shell completion through its CLI implementation (cobra). Generate the completion code for your shell by running one of the following commands:

  • grype completion <bash|zsh|fish>
  • go run main.go completion <bash|zsh|fish>

This will output a shell script to STDOUT, which can then be used as a completion script for Grype. Running one of the above commands with the
-h or --help flags will provide instructions on how to do that for your chosen shell.

Private Registry Authentication

Local Docker Credentials

When a container runtime is not present, grype can still utilize credentials configured in common credential sources (such as ~/.docker/config.json).
It will pull images from private registries using these credentials. The config file is where your credentials are stored when authenticating with private registries via some command like docker login.
For more information see the go-containerregistry documentation.

An example config.json looks something like this:

// config.json
	"auths": {
		"": {
			"username": "AzureDiamond",
			"password": "hunter2"

You can run the following command as an example. It details the mount/environment configuration a container needs to access a private registry:

docker run -v ./config.json:/config/config.json -e "DOCKER_CONFIG=/config" anchore/grype:latest <private_image>

Docker Credentials in Kubernetes

The below section shows a simple workflow on how to mount this config file as a secret into a container on kubernetes.

  1. Create a secret. The value of config.json is important. It refers to the specification detailed here.
    Below this section is the secret.yaml file that the pod configuration will consume as a volume.
    The key config.json is important. It will end up being the name of the file when mounted into the pod.

    # secret.yaml
    apiVersion: v1
    kind: Secret
      name: registry-config
      namespace: grype 
      config.json: <base64 encoded config.json>

    kubectl apply -f secret.yaml

  2. Create your pod running grype. The env DOCKER_CONFIG is important because it advertises where to look for the credential file.
    In the below example, setting DOCKER_CONFIG=/config informs grype that credentials can be found at /config/config.json.
    This is why we used config.json as the key for our secret. When mounted into containers the secrets’ key is used as the filename.
    The volumeMounts section mounts our secret to /config. The volumes section names our volume and leverages the secret we created in step one.

    # pod.yaml
    apiVersion: v1
    kind: Pod
        - image: anchore/grype:latest
          name: grype-private-registry-demo
            - name: DOCKER_CONFIG
              value: /config
          - mountPath: /config
            name: registry-config
            readOnly: true
            - <private_image>
      - name: registry-config
          secretName: registry-config

    kubectl apply -f pod.yaml

  3. The user can now run kubectl logs grype-private-registry-demo. The logs should show the grype analysis for the <private_image> provided in the pod configuration.

Using the above information, users should be able to configure private registry access without having to do so in the grype or syft configuration files.
They will also not be dependent on a docker daemon, (or some other runtime software) for registry configuration and access.


Configuration search paths:

  • .grype.yaml
  • .grype/config.yaml
  • ~/.grype.yaml
  • <XDG_CONFIG_HOME>/grype/config.yaml

Configuration options (example values are the default):

# enable/disable checking for application updates on startup
check-for-app-update: true

# same as --fail-on ; upon scanning, if a severity is found at or above the given severity then the return code will be 1
# default is unset which will skip this validation (options: negligible, low, medium, high, critical)
fail-on-severity: ''

# same as -o ; the output format of the vulnerability report (options: table, json, cyclonedx)
output: "table"

# same as -s ; the search space to look for packages (options: all-layers, squashed)
scope: "squashed"

# same as -q ; suppress all output (except for the vulnerability list)
quiet: false

# same as --file; write output report to a file (default is to write to stdout)
file: ""

# a list of globs to exclude from scanning. same as --exclude ; for example:
# exclude:
#   - '/etc/**'
#   - './out/**/*.json'

  # check for database updates on execution
  auto-update: true

  # location to write the vulnerability database cache
  cache-dir: "$XDG_CACHE_HOME/grype/db"

  # URL of the vulnerability database
  update-url: ""

# options when pulling directly from a registry via the "registry:" scheme
  # skip TLS verification when communicating with the registry
  insecure-skip-tls-verify: false
  # use http instead of https when connecting to the registry
  insecure-use-http: false

  # credentials for specific registries
    - # the URL to the registry (e.g. "", "localhost:5000", etc.)
      authority: ""
      username: ""
      password: ""
      # note: token and username/password are mutually exclusive
      token: ""
    - ... # note, more credentials can be provided via config file only

  # location to write the log file (default is not to have a log file)
  file: ""

  # the log level; note: detailed logging suppress the ETUI
  level: "error"

  # use structured logging
  structured: false

Future plans

The following areas of potential development are currently being investigated:

  • Support for allowlist, package mapping
  • Accept alternative SBOM formats (CycloneDX, SPDX) as input


View Github