GitHub's Policy as Code project is designed to allow users and organizations to configure their Risk threshold for security issues reported by GitHub Advanced Security Platform. The main goal is to help make sure that before publishing your application to productions, development, etc. you can validate if the application has any security issues that need to be addressed.
- Highly Configurable
- Re-usable across repositories
- Supports all GitHub Advanced Security Features
- Supports GitHub Enterprise Cloud or Server (see supported features list)
- Pull Request Summary
- Python +3.9
Here is how you can quickly setup policy-as-code.
Tip
Checkout the GitHub Actions Policy as Code Examples
# Policy as Code
- name: Advance Security Policy as Code
uses: advanced-security/[email protected]Warning
The GitHub Action does not install Python on the runner. Please checkout at the actions/setup-python Action
The Policy as Code project is a self-contained Python based CLI tool.
Note
All of the Dependencies for Policy as Code are vendored into this repository
Bash / Zsh:
git clone --branch "v2.9.0" https://github.com/advanced-security/policy-as-code.git && cd ./policy-as-code
./policy-as-code --helpPowershell:
git clone --branch "v2.9.0" https://github.com/advanced-security/policy-as-code.git
cd policy-as-code
.\policy-as-code.ps1 --helpTip
Checkout the samples of how to use / run the cli with examples.
For Policy as Code to work correctly, you need to have the following permissions for the different features:
- [required] Repository Permissions
security_events: readcontent: readpull-requests: write- Policy as Code Pull Request Summary
- [optional] Policy Repository
content: readto be able to clone external sources of the policies
# workflow or job level
permissions:
contents: read
security-events: read
# pull request summaries
pull-requests: write| Feature | github.com (Br/PR) | enterprise server (Br/PR) | Description |
|---|---|---|---|
| Code Scanning | ✅ / ✅ | ✅ / ✅ | Code Scanning is a code analysis tool that scans your code for security vulnerabilities and coding errors. |
| Secret Scanning | ✅ / ✅ | ✅ / ✅ | Secret Scanning is a code analysis tool that scans your code for secrets. |
| Dependabot | ✅ / ✅ | ❌ [2] / ✅ [3] | Dependabot is a tool that automates discovery and remediation of vulnerabilities in your dependencies. |
| Dependencies | ✅ / ✅ | ❌ [2] / ✅ [3] | Dependencies is a tool that scans your code for dependencies. |
| Licensing | ✅ / ✅ | ❌ [2] / ✅ [3] | Licensing is a tool that scans your code for licensing issues. |
Notes:
- Br/PR = Branches / Pull Requests
⚠️ GraphQL API not supported by GitHub Enterprise Server as of3.7- ✅ Supported as of GHES
3.6
Here is an example of using a simple yet cross-organization using Policy as Code:
# Compliance
- name: Advance Security Policy as Code
uses: advanced-security/[email protected]
with:
# The owner/repo of where the policy is stored
policy: GeekMasher/security-queries
# The local (within the workspace) or repository
policy-path: policies/default.yml
# The branch you want to target
policy-branch: mainThe Policy as Code configuration file is very simple yet powerful allowing a user to define 4 types of rules per technologies you want to use.
# This is the technology you want to write a rule for
licensing:
# The four main rules types to do everything you need to do for all things
# compliance
# Warnings will always occur if the rule applies and continues executing to
# other rules.
warnings:
ids:
- Other
- NA
# Ignores are run next so if an ignored rule is hit that matches the level,
# it will be skipped
ignores:
ids:
- MIT License
# Conditions will only trigger and raise an error when an exact match is hit
conditions:
ids:
- GPL-2.0
names:
- tunnel-agent
# The simplest and ultimate rule which checks the severity of the alert and
# reports an issue if the level matches or higher (see PaC Levels for more info)
level: errorThere are many different levels of severities with the addition of all and none (self explanatory).
When a level is selected like for example error, all higher level severities (critical and high in this example) will also be added.
- critical
- high
- error
- medium
- moderate
- low
- warning
- notesFor each rule you can choose either or both of the two different criteria's matches; ids and names
You can also use imports to side load data from other files to supplement the data already in the rule block
codescanning:
conditions:
# When the `ids` of the technologies/tool alert matches any one of the ID's in
# the list specified, the rule will the triggered and report the alert.
ids:
# In this example case, the CodeQL rule ID below will always be reported if
# present event if the severity is low or even note.
- js/sql-injection
# Side note: Check to see what different tools consider id's verses names,
# for example `licensing` considers the "Licence" name itself as the id
# while the name of the package/library as the "name"
# `names` allows you to specify the names of alerts or packages.
names:
- "Missing rate limiting"
# The `imports` allows you to supplement your existing data with a list
# from a file on the system.
imports:
ids: "path/to/ids/supplement/file.txt"
names: "path/to/names/supplement/file.txt"For both types of criteria matching you can use wildcards to easily match requirements in a quicker way.
The matching is done using a Unix shell-style wildcards module called fnmatch which supports * for matching everything.
codescanning:
conditions:
ids:
- "*/sql-injection"The feature allows a user to define a time frame to which a security alert/vulnerability of a certain severity has before the alert triggered a violation in the Action.
By default, if this section is not defined in any part of the policy then no checks are done. Existing policy files should act the same without the new section.
general:
# All other blocks will be inheriting the remediate section if they don't have
# their own defined.
remediate:
# Only `error`'s and above have got 7 days to remediate according to the
# policy. Any time before that, nothing will occur and post the remediation
# time frame the alert will be raised.
error: 7
codescanning:
# the `codescanning` block will inherit the `general` block
# ...
dependabot:
remediate:
# high and critical security issues
high: 7
# moderate security issues
moderate: 30
# all other security issues
all: 90
secretscanning:
remediate:
# All secrets by default are set to 'critical' severity so only `critical`
# or `all` will work
critical: 7Some things to consider when using imports:
- Imports appending to existing lists and do not replace a previously generated list.
- Imports are relative to:
Working DirectoryGitHub Action / CLI directoryCloned Repository Directory
- Imports are only allowed from a number of predefined paths to prevent loading data on the system (AKA, path traversal).
- @GeekMasher - Author / Core Maintainer
Please create GitHub Issues if there are bugs or feature requests.
This project uses Sematic Versioning (v2) and with major releases, breaking changes will occur.
This project is licensed under the terms of the MIT open source license. Please refer to MIT for the full terms.
![@dependabot[bot]](https://avatars.githubusercontent.com/in/29110?s=64&v=4){kind=small}
