The CVE Binary Tool is a free, open source tool to help you find known vulnerabilities in software, using data from the National Vulnerability Database (NVD) list of Common Vulnerabilities and Exposures (CVEs).
The tool has two main modes of operation:
- A binary scanner which helps you determine which packages may have been included as part of a piece of software. There are around 100 checkers which focus on common, vulnerable open source components such as openssl, libpng, libxml2 and expat.
- Tools for scanning known component lists in various formats, including .csv, several linux distribution package lists, language specific package scanners and several Software Bill of Materials (SBOM) formats.
It is intended to be used as part of your continuous integration system to enable regular vulnerability scanning and give you early warning of known issues in your supply chain.
For more details, see our documentation or this quickstart guide
- CVE Binary Tool quick start / README
CVE Binary Tool can be installed using pip:
pip install cve-bin-tool
You can also do pip install --user -e .
to install a local copy which is useful if you're trying the latest code from
the cve-bin-tool github or doing development. The Contributor Documentation covers how to set up for local development in more detail.
To run the binary scanner on a directory or file:
cve-bin-tool <directory/file>
Note that this option will also use any language specific checkers to find known vulnerabilities in components.
To scan a comma-delimited (CSV) or JSON file which lists dependencies and versions:
cve-bin-tool --input-file <filename>
Note that the --input-file
option can also be used to add extra triage data like remarks, comments etc. while scanning a directory so that output will reflect this triage data and you can save time of re-triaging (Usage: cve-bin-tool -i=test.csv /path/to/scan
).
A VEX file (which may be created using the --vex
command line option) can also be used as a triage file. A VEX file
is detected if the file suffix is '.vex'.
To scan a software bill of materials file (SBOM):
cve-bin-tool --sbom <sbom_filetype> --sbom-file <sbom_filename>
Valid SBOM types are SPDX, CycloneDX, and SWID.
Specifying the --offline
option when running a scan ensures that cve-bin-tool doesn't attempt to download the latest database files or to check for a newer version of the tool.
Note that you will need to obtain a copy of the vulnerability data before the tool can run in offline mode. The offline how-to guide contains more information on how to set up your database.
The CVE Binary Tool provides console-based output by default. If you wish to provide another format, you can specify this and a filename on the command line using --format
. The valid formats are CSV, JSON, console, HTML and PDF. The output filename can be specified using the --output-file
flag.
You can also specify multiple output formats by using comma (',') as separator:
cve-bin-tool file -f csv,json,html -o report
Note: Please don't use spaces between comma (',') and the output formats.
The reported vulnerabilities can additionally be reported in the
Vulnerability Exchange (VEX) format by specifying --vex
command line option.
The generated VEX file can then be used as an --input-file
to support
a triage process.
If you wish to use PDF support, you will need to install the reportlab
library separately.
If you intend to use PDF support when you install cve-bin-tool you can specify it and report lab will be installed as part of the cve-bin-tool install:
pip install cve-bin-tool[PDF]
If you've already installed cve-bin-tool you can add reportlab after the fact using pip:
pip install --upgrade reportlab
Note that reportlab was taken out of the default cve-bin-tool install because it has a known CVE associated with it (CVE-2020-28463). The cve-bin-tool code uses the recommended mitigations to limit which resources added to PDFs, as well as additional input validation. This is a bit of a strange CVE because it describes core functionality of PDFs: external items, such as images, can be embedded in them, and thus anyone viewing a PDF could load an external image (similar to how viewing a web page can trigger external loads). There's no inherent "fix" for that, only mitigations where users of the library must ensure only expected items are added to PDFs at the time of generation.
Since users may not want to have software installed with an open, unfixable CVE associated with it, we've opted to make PDF support only available to users who have installed the library themselves. Once the library is installed, the PDF report option will function.
Usage:
cve-bin-tool <directory/file to scan>
options:
-h, --help show this help message and exit
-e EXCLUDE, --exclude EXCLUDE
Comma separated Exclude directory path
-V, --version show program's version number and exit
--disable-version-check
skips checking for a new version
--disable-validation-check
skips checking xml files against schema
--offline operate in offline mode
--detailed display detailed report
CVE Data Download:
Arguments related to data sources and Cache Configuration
-n {api,json}, --nvd {api,json}
choose method for getting CVE lists from NVD
-u {now,daily,never,latest}, --update {now,daily,never,latest}
update schedule for data sources and exploits database (default: daily)
--nvd-api-key NVD_API_KEY
specify NVD API key (used to improve NVD rate limit)
-d {NVD,OSV} [{NVD,OSV} ...], --disable-data-source {NVD,OSV} [{NVD,OSV} ...]
specify data sources that should be disabled
Input:
directory directory to scan
-i INPUT_FILE, --input-file INPUT_FILE
provide input filename
-C CONFIG, --config CONFIG
provide config file
-L PACKAGE_LIST, --package-list PACKAGE_LIST
provide package list
--sbom {spdx,cyclonedx,swid}
specify type of software bill of materials (sbom) (default: spdx)
--sbom-file SBOM_FILE
provide sbom filename
Output:
-q, --quiet suppress output
-l {debug,info,warning,error,critical}, --log {debug,info,warning,error,critical}
log level (default: info)
-o OUTPUT_FILE, --output-file OUTPUT_FILE
provide output filename (default: output to stdout)
--html-theme HTML_THEME
provide custom theme directory for HTML Report
-f {csv,json,console,html,pdf}, --format {csv,json,console,html,pdf}
update output format (default: console)
specify multiple output formats by using comma (',') as a separator
note: don't use spaces between comma (',') and the output formats.
-c CVSS, --cvss CVSS minimum CVSS score (as integer in range 0 to 10) to report (default: 0)
-S {low,medium,high,critical}, --severity {low,medium,high,critical}
minimum CVE severity to report (default: low)
--report Produces a report even if there are no CVE for the respective output format
-A [<distro_name>-<distro_version_name>], --available-fix [<distro_name>-<distro_version_name>]
Lists available fixes of the package from Linux distribution
-b [<distro_name>-<distro_version_name>], --backport-fix [<distro_name>-<distro_version_name>]
Lists backported fixes if available from Linux distribution
--affected-versions Lists versions of product affected by a given CVE (to facilitate upgrades)
--vex VEX Provide vulnerability exchange (vex) filename
Merge Report:
Arguments related to Intermediate and Merged Reports
-a [APPEND], --append [APPEND]
save output as intermediate report in json format
-t TAG, --tag TAG add a unique tag to differentiate between multiple intermediate reports
-m MERGE, --merge MERGE
comma separated intermediate reports path for merging
-F FILTER, --filter FILTER
comma separated tag string for filtering intermediate reports
Checkers:
-s SKIPS, --skips SKIPS
comma-separated list of checkers to disable
-r RUNS, --runs RUNS comma-separated list of checkers to enable
Database Management:
--export EXPORT export database filename
--import IMPORT import database filename
Exploits:
--exploits check for exploits from found cves
Deprecated:
-x, --extract autoextract compressed files
CVE Binary Tool autoextracts all compressed files by default now
For further information about all of these options, please see the CVE Binary Tool user manual.
Note: For backward compatibility, we still support
csv2cve
command for producing CVEs from csv but we recommend using the--input-file
command going forwards.
-L
or --package-list
option runs a CVE scan on installed packages listed in a package list. It takes a python package list (requirements.txt) or a package list of packages of systems that has dpkg, pacman or rpm package manager as an input for the scan. This option is much faster and detects more CVEs than the default method of scanning binaries.
You can get a package list of all installed packages in
- a system using dpkg package manager by running
dpkg-query -W -f '${binary:Package}\n' > pkg-list.txt
- a system using pacman package manager by running
pacman -Qqe > pkg-list.txt
- a system using rpm package manager by running
rpm -qa --queryformat '%{NAME}\n' > pkg-list.txt
in the terminal and provide it as an input by running cve-bin-tool -L pkg-list.txt
for a full package scan.
You can use --config
option to provide configuration file for the tool. You can still override options specified in config file with command line arguments. See our sample config files in the
test/config
If you want to integrate cve-bin-tool as a part of your github action pipeline. You can checkout our example github action.
The following data sources are used to get CVE data to find CVEs for a package:
This data source consists of majority of the CVE entries and is essential to provide vendor data for other data sources such as OSV, therefore downloading CVE data from it cannot be disabled, --disable-data-source "NVD"
only disables CVEs from displaying in output.
This data source is based on the OSV schema from google, and consists of CVEs from different ecosystems that might not be covered by NVD. NVD is given priority if there are duplicate CVEs as some CVEs from OSV may not contain CVSS scores. Using OSV will increase number of CVEs and time taken to update the database but searching database for vulnerabilities will have similar performance.
Gitlab Advisory Database (GAD)
This data source consists of security advisories used by the GitLab dependency scanner. Amount of CVEs added from this data source is similar to OSV.
The following checkers are available for finding components in binary files:
Available checkers | ||||||
---|---|---|---|---|---|---|
accountsservice | acpid | apache_http_server | apcupsd | avahi | bash | bind |
binutils | bolt | bubblewrap | busybox | bzip2 | collectd | commons_compress |
cronie | cryptsetup | cups | curl | dbus | dhcpcd | dnsmasq |
dovecot | dpkg | enscript | expat | fastd | ffmpeg | freeradius |
ftp | gcc | gimp | glibc | gnomeshell | gnupg | gnutls |
gpgme | gstreamer | gupnp | haproxy | haserl | hdf5 | hostapd |
hunspell | i2pd | icecast | icu | irssi | iucode_tool | jacksondatabind |
kbd | keepalived | kerberos | kexectools | libarchive | libbpg | libdb |
libebml | libgcrypt | libical | libjpeg_turbo | liblas | libnss | librsvg |
libsamplerate | libseccomp | libsndfile | libsolv | libsoup | libsrtp | libssh |
libssh2 | libtiff | libupnp | libvirt | libvncserver | libvorbis | libxslt |
lighttpd | lldpd | logrotate | lua | luajit | mariadb | mdadm |
memcached | minicom | miniupnpd | motion | mtr | mutt | mysql |
nano | ncurses | nessus | netpbm | nginx | node | ntp |
open_vm_tools | openafs | openjpeg | openldap | openssh | openssl | openswan |
openvpn | p7zip | patch | pcsc_lite | pigz | png | polarssl_fedora |
poppler | postgresql | privoxy | pspp | python | qt | radare2 |
rsyslog | rust | samba | sane_backends | sqlite | strongswan | subversion |
sudo | syslogng | systemd | tcpdump | tinyproxy | trousers | varnish |
webkitgtk | wireshark | wpa_supplicant | xerces | xml2 | zlib | zsh |
All the checkers can be found in the checkers directory, as can the instructions on how to add a new checker. Support for new checkers can be requested via GitHub issues.
A number of checkers are available for finding vulnerable components in specific language packages.
The scanner examines the pom.xml
file within a Java package archive to identify Java components. The package names and versions within the archive are used to search the database for vulnerabilities.
JAR, WAR and EAR archives are supported.
The scanner examines the package-lock.json
file within a javascript application
to identify components. The package names and versions are used to search the database for vulnerabilities.
The scanner examines the Cargo.lock
file which is created by cargo to manage the dependencies of the project with their specific versions. The package names and versions are used to search the database for vulnerabilities.
The scanner examines the Gemfile.lock
file which is created by bundle to manage the dependencies of the project with their specific versions. The package names and versions are used to search the database for vulnerabilities.
The scanner examines the renv.lock
file which is created by renv to manage the dependencies of the project with their specific versions. The package names and versions are used to search the database for vulnerabilities.
The scanner examines the go.mod
file which is created by mod to manage the dependencies of the project with their specific versions. The package names and versions are used to search the database for vulnerabilities.
The scanner examines the PKG-INFO
and METADATA
files for an installed Python package to extract the component name and version which
are used to search the database for vulnerabilities.
The tool supports the scanning of the contents of any Wheel package files (indicated with a file extension of .whl) and egg package files (indicated with a file extension of .egg).
The --package-list
option can be used with a Python dependencies file requirements.txt
to find the vulnerabilities in the list of components.
This scanner does not attempt to exploit issues or examine the code in greater detail; it only looks for library signatures and version numbers. As such, it cannot tell if someone has backported fixes to a vulnerable version, and it will not work if library or version information was intentionally obfuscated.
This tool is meant to be used as a quick-to-run, easily-automatable check in a non-malicious environment so that developers can be made aware of old libraries with security issues that have been compiled into their binaries.
The tool does not guarantee that any vulnerabilities reported are actually present or exploitable, neither is it able to find all present vulnerabilities with a guarantee.
Users can add triage information to reports to mark issues as false positives, false negatives, indicate that the risk has been mitigated by configuration/usage changes, and so on.
Triage details can be re-used on other projects so, for example, triage on a Linux base image could be applied to multiple containers using that image.
For more information and usage of triage information with the tool kindly have a look here.
If you are using the binary scanner capabilities, be aware that we only have a limited number of binary checkers (see table above) so we can only detect those libraries. Contributions of new checkers are always welcome! You can also use an alternate way to detect components (for example, a bill of materials tool such as tern) and then use the resulting list as input to cve-bin-tool to get a more comprehensive vulnerability list.
The tool uses a vulnerability database in order to detect the present vulnerabilities, in case the database is not frequently updated (specially if the tool is used in offline mode), the tool would be unable to detect any newly discovered vulnerabilities. Hence it is highly advised to keep the database updated.
To use the auto-extractor, you may need the following utilities depending on the type of file you need to extract. The utilities below are required to run the full test suite on Linux:
file
strings
tar
unzip
rpm2cpio
cpio
ar
cabextract
Most of these are installed by default on many Linux systems, but cabextract
and
rpm2cpio
in particular might need to be installed.
On windows systems, you may need:
ar
7z
Expand
pdftotext
Windows has ar
and Expand
installed by default, but 7z
in particular might need to be installed.
If you want to run our test-suite or scan a zstd compressed file, We recommend installing this 7-zip-zstd
fork of 7zip. We are currently using 7z
for extracting jar
, apk
, msi
, exe
and rpm
files.
If you get an error about building libraries when you try to install from pip, you may need to install the Windows build tools. The Windows build tools are available for free from https://visualstudio.microsoft.com/visual-cpp-build-tools/
If you get an error while installing brotlipy on Windows, installing the compiler above should fix it.
pdftotext
is required for running tests. (users of cve-bin-tool may not need it, developers likely will.) The best approach to install it on Windows involves using conda (click here for further instructions).
You can check our CI configuration to see what versions of python we're explicitly testing.
Bugs and feature requests can be made via GitHub issues. Be aware that these issues are not private, so take care when providing output to make sure you are not disclosing security issues in other products.
Pull requests are also welcome via git.
- New contributors should read the contributor guide to get started.
- Folk who already have experience contributing to open source projects may not need the full guide but should still use the pull request checklist to make things easy for everyone.
CVE Binary Tool contributors are asked to adhere to the Python Community Code of Conduct. Please contact Terri if you have concerns or questions relating to this code of conduct.
Security issues with the tool itself can be reported to Intel's security incident response team via https://intel.com/security.
If in the course of using this tool you discover a security issue with someone else's code, please disclose responsibly to the appropriate party.