. . . Skjold /skjɔl/
,-. | , . ,-. | ,-|
`-. |< | | | | | | Security audit python project dependencies
`-' ' ` | `-' `' `-´ against several security advisory databases.
`'
Introduction
It currently supports fetching advisories from the following sources:
| Source | Name | Notes |
|---|---|---|
| GitHub Advisory Database | github |
|
| PyUP.io safety-db | pyup |
|
| GitLab gemnasium-db | gemnasium |
|
| PYPA Advisory Database | pypa |
Experimental! Only supports ECOSYSTEM and SEMVER! |
| OSV.dev Database | osv |
Experimental! Only supports ECOSYSTEM and SEMVER!Sends package information to OSV.dev API. |
No source is enabled by default! Individual sources can be enabled by setting sources list (see Configuration). There is (currently) no de-duplication meaning that using all of them could result in a lot of duplicates.
Motivation
Skjold was initially created for myself to replace safety. Which appears to no longer receive monthly updates (see pyupio/safety-db #2282). I wanted something I can run locally and use for my local or private projects/scripts.
I currently also use it during CI builds and before deploying/publishing containers or packages.
Installation
skjold can be installed from either PyPI or directly from Github using pip:
pip install skjold # Install from PyPI
pip install git+https://github.com/twu/[email protected] # Install from Github
This should provide a script named skjold that can then be invoked. See Usage.
Usage
$ pip list --format=freeze | skjold -v audit --sources gemnasium -
When running audit one can either provide a path to a frozen requirements.txt, a poetry.lock or a Pipfile.lock file. Alternatively, dependencies can also be passed in via stdin (formatted as package==version).
skjold will maintain a local cache (under cache_dir) that will expire automatically after cache_expires has passed. The cache_dir and cache_expires can be adjusted by setting them in tools.skjold section of the projects pyproject.toml (see Configuration for more details). The cache_dirwill be created automatically, and by default unless otherwise specified will be located under $HOME/.skjold/cache.
For further options please read skjold --help and/or skjold audit --help.
Examples
All examples involving github assume that SKJOLD_GITHUB_API_TOKEN is already set (see Github).
# Using pip list. Checking against GitHub only.
$ pip list --format=freeze | skjold audit -s github -
# Be verbose. Read directly from supported formats.
$ skjold -v audit requirements.txt
$ skjold -v audit poetry.lock
$ skjold -v audit Pipenv.lock
# Using poetry.
$ poetry export -f requirements.txt | skjold audit -s github -s gemnasium -s pyup -
# Using poetry, format output as json and pass it on to jq for additional filtering.
$ poetry export -f requirements.txt | skjold audit -o json -s github - | jq '.[0]'
# Using Pipenv, checking against Github
$ pipenv run pip list --format=freeze | skjold audit -s github -
# Checking a single package via stdin against Github and format findings as json.
$ echo "urllib3==1.23" | skjold audit -o json -r -s github -
[
{
"severity": "HIGH",
"name": "urllib3",
"version": "1.23",
"versions": "<1.24.2",
"source": "github",
"summary": "High severity vulnerability that affects urllib3",
"references": [
"https://nvd.nist.gov/vuln/detail/CVE-2019-11324"
],
"url": "https://github.com/advisories/GHSA-mh33-7rrq-662w"
}
]
# Checking a single package via stdin against Gemnasium and report findings (`-o cli`).
$ echo "urllib3==1.23" | skjold audit -o cli -r -s gemnasium -
urllib3==1.23 (<=1.24.2) via gemnasium
CRLF injection. In the urllib3 library for Python, CRLF injection is possible
if the attacker controls the request parameter.
https://nvd.nist.gov/vuln/detail/CVE-2019-11236
--
urllib3==1.23 (<1.24.2) via gemnasium
Weak Authentication Caused By Improper Certificate Validation. The urllib3
library for Python mishandles certain cases where the desired set of CA
certificates is different from the OS store of CA certificates, which results
in SSL connections succeeding in situations where a verification failure is the
correct outcome. This is related to use of the `ssl_context`, `ca_certs`, or
`ca_certs_dir` argument.
https://nvd.nist.gov/vuln/detail/CVE-2019-11324
--
urllib3==1.23 (<1.25.9) via gemnasium
Injection Vulnerability. urllib3 allows CRLF injection if the attacker controls
the HTTP request method, as demonstrated by inserting `CR` and `LF` control
characters in the first argument of `putrequest()`. NOTE: this is similar to
CVE-2020-26116.
https://nvd.nist.gov/vuln/detail/CVE-2020-26137
--
Ignore Findings
Findings can be ignored either by manually adding an entry using the sources identifier to a file named .skjoldignore (See Example) or by using in the CLI. Below are a few possible usage examples.
# Ignore PYSEC-2020-148 finding from PyPA source until a certain date with a specific reason.
$ skjold ignore urllib3 PYSEC-2020-148 --reason "Very good reason." --expires "2021-01-01T00:00:00+00:00"
Ignore urllib3 in PYSEC-2020-148 until 2021-01-01 00:00:00+00:00?
Very good reason.
--
Add to '.skjoldignore'? [y/N]: y
# Ignore PYSEC-2020-148 finding from PyPA source for 7 days with "No immediate remediation." reason.
$ skjold ignore urllib3 PYSEC-2020-148
Ignore urllib3 in PYSEC-2020-148 until ...?
No immediate remediation.
--
Add to '.skjoldignore'? [y/N]: y
# Audit `poetry.lock` using a custom `.skjoldignore` file location via `ENV`...
$ SKJOLD_IGNORE_FILE=
skjold audit -s pyup poetry.lock
# ... or using -i/--ignore-file
$ skjold audit -s pyup -i
poetry.lock
Configuration
skjold can read its configuration from the tools.skjold section of a projects pyproject.toml. Arguments specified via the command-line should take precedence over any configured or default value.
[tool.skjold]
sources = ["github", "pyup", "gemnasium"] # Sources to check against.
report_only = true # Report only, always exit with zero.
report_format = 'json' # Output findings as `json`. Default is 'cli'.
cache_dir = '.skjold_cache' # Cache location (default: `~/.skjold/cache`).
cache_expires = 86400 # Cache max. age.
ignore_file = '.skjoldignore' # Ignorefile location (default `.skjoldignore`).
verbose = true # Be verbose.
To take a look at the current configuration / defaults run:
$ skjold config
sources: ['pyup', 'github', 'gemnasium']
report_only: True
report_format: json
verbose: False
cache_dir: .skjold_cache
cache_expires: 86400
ignore_file = '.skjoldignore'
Github
For the github source to work you'll need to provide a Github API Token via an ENV variable named SKJOLD_GITHUB_API_TOKEN. You can create a new Github Access Token here. You do not have to give it any permissions as it is only required to query the GitHub GraphQL API v4 API.
Version Control Integration
To use skjold with the excellent pre-commit framework add the following to the projects .pre-commit-config.yaml after installation.
repos:
- repo: https://github.com/twu/skjold
rev: vX.X.X
hooks:
- id: skjold
verbose: true # Important if used with `report_only`, see below.
After running pre-commit install the hook should be good to go. To configure skjold in this scenario I recommend adding the entire configuration to the projects pyproject.toml instead of manipulating the hook args. See this projects pyproject.toml for an example.
Important!: When using
skjoldas apre-commit-hook it only gets triggered if you want to commit changed dependency files (e.g.Pipenv.lock,poetry.lock,requirements.txt,...). It will not continuously check your dependencies on every commit!
You could run pre-commit run skjold --all-files manually in your workflow/scripts or run skjold manually. If you have a better solution please let me know!
Important!: If you use
report_onlyin any way make sure that you addverbose: trueto your hook configuration otherwisepre-commitwon't show you any output since the hook is always returning with a zero exit code due toreport_onlybeing set!
Contributing
Pull requests are welcome! For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.
2 Feb 15, 2022
96 Jan 02, 2023
16 Dec 30, 2022
33 Sep 28, 2022
38 Aug 09, 2022
1 Nov 25, 2021
22 Dec 04, 2022
85 Dec 02, 2022
87 Dec 29, 2021
45 Dec 20, 2022
1 Jan 11, 2022
53 Nov 01, 2022
4 Oct 10, 2022
1 Jan 02, 2022
44 Nov 15, 2022
17 Aug 14, 2022
2 Jun 06, 2022
772 Jan 05, 2023
4 Mar 07, 2022
4.8k Dec 31, 2022