Installation
Documents how to install and use the Safety CLI tool to check Python dependencies for known security vulnerabilities.
What this file does
Documents how to install and use the Safety CLI tool to check Python dependencies for known security vulnerabilities.
When to use it
- You want to audit your Python project's dependencies for security issues
- You need to integrate vulnerability scanning into a CI pipeline
- You prefer a CLI tool that reads requirements files or pip freeze output
- You want to run safety checks inside a Docker container
Assumes this stack
Safety checks your installed dependencies for known security vulnerabilities.
By default it uses the open Python vulnerability database Safety DB,
but can be upgraded to use pyup.io's Safety API using the --key option.
Installation
Install safety with pip
pip install safety
Usage
To check your currently selected virtual environment for dependencies with known security vulnerabilites, run:
safety check
You should get a report similar to this:
╒══════════════════════════════════════════════════════════════════════════════╕
│ │
│ /$$$$$$ /$$ │
│ /$$__ $$ | $$ │
│ /$$$$$$$ /$$$$$$ | $$ \__//$$$$$$ /$$$$$$ /$$ /$$ │
│ /$$_____/ |____ $$| $$$$ /$$__ $$|_ $$_/ | $$ | $$ │
│ | $$$$$$ /$$$$$$$| $$_/ | $$$$$$$$ | $$ | $$ | $$ │
│ \____ $$ /$$__ $$| $$ | $$_____/ | $$ /$$| $$ | $$ │
│ /$$$$$$$/| $$$$$$$| $$ | $$$$$$$ | $$$$/| $$$$$$$ │
│ |_______/ \_______/|__/ \_______/ \___/ \____ $$ │
│ /$$ | $$ │
│ | $$$$$$/ │
│ by pyup.io \______/ │
│ │
╞══════════════════════════════════════════════════════════════════════════════╡
│ REPORT │
╞══════════════════════════════════════════════════════════════════════════════╡
│ No known security vulnerabilities found. │
╘══════════════════════════════════════════════════════════════════════════════╛
Now, let's install something insecure:
pip install insecure-package
Yeah, you can really install that.
Run safety check again:
╒══════════════════════════════════════════════════════════════════════════════╕
│ │
│ /$$$$$$ /$$ │
│ /$$__ $$ | $$ │
│ /$$$$$$$ /$$$$$$ | $$ \__//$$$$$$ /$$$$$$ /$$ /$$ │
│ /$$_____/ |____ $$| $$$$ /$$__ $$|_ $$_/ | $$ | $$ │
│ | $$$$$$ /$$$$$$$| $$_/ | $$$$$$$$ | $$ | $$ | $$ │
│ \____ $$ /$$__ $$| $$ | $$_____/ | $$ /$$| $$ | $$ │
│ /$$$$$$$/| $$$$$$$| $$ | $$$$$$$ | $$$$/| $$$$$$$ │
│ |_______/ \_______/|__/ \_______/ \___/ \____ $$ │
│ /$$ | $$ │
│ | $$$$$$/ │
│ by pyup.io \______/ │
│ │
╞══════════════════════════════════════════════════════════════════════════════╡
│ REPORT │
╞══════════════════════════╤═══════════════╤═══════════════════╤═══════════════╡
│ package │ installed │ affected │ source │
╞══════════════════════════╧═══════════════╧═══════════════════╧═══════════════╡
│ insecure-package │ 0.1.0 │ <0.2.0 │ changelog │
╘══════════════════════════╧═══════════════╧═══════════════════╧═══════════════╛
Examples
Read requirement files
Just like pip, Safety is able to read local requirement files:
safety check -r requirements.txt
Read from stdin
Safety is also able to read from stdin with the --stdin flag set.
To check a local requirements file, run:
cat requirements.txt | safety check --stdin
or the output of pip freeze:
pip freeze | safety check --stdin
or to check a single package:
echo "insecure-package==0.1" | safety check --stdin
For more examples, take a look at the options section.
Using Safety in Docker
Safety can be easily executed as Docker container. To build the container just execute:
docker build -t safety-docker .
The container can be used just as described in the examples section.
echo "insecure-package==0.1" | docker run -i --rm safety-docker safety check --stdin
cat requirements_dev.txt | docker run -i --rm safety-docker safety check --stdin
Using Safety with a CI service
Safety works great in your CI pipeline. It returns a non-zero exit status if it finds a vulnerability.
Run it before or after your tests. If Safety finds something, your tests will fail.
Travis
install:
- pip install safety
script:
- safety check
Deep GitHub Integration
If you are looking for a deep integration with your GitHub repositories: Safety is available as a part of pyup.io, called Safety CI. Safety CI checks your commits and pull requests for dependencies with known security vulnerabilities and displays a status on GitHub.
Using Safety in production
Safety is free and open source (MIT Licensed). The underlying open vulnerability database is updated once per month.
To get access to all vulnerabilites as soon as they are added, you need a Safety API key that comes with a paid pyup.io account, starting at $99 for organizations.
Options
--key
API Key for pyup.io's vulnerability database. Can be set as SAFETY_API_KEY environment variable.
Example
safety check --key=12345-ABCDEFGH
--db
Path to a directory with a local vulnerability database including insecure.json and insecure_full.json
Example
safety check --db=/home/safety-db/data
--proxy-host
Proxy host IP or DNS
--proxy-port
Proxy port number
--proxy-protocol
Proxy protocol (https or http)
--json
Output vulnerabilities in JSON format.
Example
safety check --json
[
[
"django",
"<1.2.2",
"1.2",
"Cross-site scripting (XSS) vulnerability in Django 1.2.x before 1.2.2 allows remote attackers to inject arbitrary web script or HTML via a csrfmiddlewaretoken (aka csrf_token) cookie.",
"25701"
]
]
--full-report
Full reports include a security advisory (if available).
Example
safety check --full-report
╒══════════════════════════════════════════════════════════════════════════════╕
│ │
│ /$$$$$$ /$$ │
│ /$$__ $$ | $$ │
│ /$$$$$$$ /$$$$$$ | $$ \__//$$$$$$ /$$$$$$ /$$ /$$ │
│ /$$_____/ |____ $$| $$$$ /$$__ $$|_ $$_/ | $$ | $$ │
│ | $$$$$$ /$$$$$$$| $$_/ | $$$$$$$$ | $$ | $$ | $$ │
│ \____ $$ /$$__ $$| $$ | $$_____/ | $$ /$$| $$ | $$ │
│ /$$$$$$$/| $$$$$$$| $$ | $$$$$$$ | $$$$/| $$$$$$$ │
│ |_______/ \_______/|__/ \_______/ \___/ \____ $$ │
│ /$$ | $$ │
│ | $$$$$$/ │
│ by pyup.io \______/ │
│ │
╞══════════════════════════════════════════════════════════════════════════════╡
│ REPORT │
╞════════════════════════════╤═══════════╤══════════════════════════╤══════════╡
│ package │ installed │ affected │ ID │
╞════════════════════════════╧═══════════╧══════════════════════════╧══════════╡
│ django │ 1.2 │ <1.2.2 │ 25701 │
╞══════════════════════════════════════════════════════════════════════════════╡
│ Cross-site scripting (XSS) vulnerability in Django 1.2.x before 1.2.2 allows │
│ remote attackers to inject arbitrary web script or HTML via a csrfmiddlewar │
│ etoken (aka csrf_token) cookie. │
╘══════════════════════════════════════════════════════════════════════════════╛
--bare
Output vulnerable packages only. Useful in combination with other tools.
Example
safety check --bare
cryptography django
--cache
Cache requests to the vulnerability database locally for 2 hours.
Example
safety check --cache
--stdin
Read input from stdin.
Example
cat requirements.txt | safety check --stdin
pip freeze | safety check --stdin
echo "insecure-package==0.1" | safety check --stdin
--file, -r
Read input from one (or multiple) requirement files.
Example
safety check -r requirements.txt
safety check --file=requirements.txt
safety check -r req_dev.txt -r req_prod.txt
--ignore, -i
Ignore one (or multiple) vulnerabilities by ID
Example
safety check -i 1234
safety check --ignore=1234
safety check -i 1234 -i 4567 -i 89101
--output, -o
Save the report to a file
Example
safety check -o insecure_report.txt
safety check --output --json insecure_report.json
Review
If you save the report in JSON format you can review in the report format again.
Options
--file, -f (REQUIRED)
Read an insecure report.
Example
safety check -f insecure.json
safety check --file=insecure.json
--full-report
Full reports include a security advisory (if available).
Example
safety review -r insecure.json --full-report
╒══════════════════════════════════════════════════════════════════════════════╕
│ │
│ /$$$$$$ /$$ │
│ /$$__ $$ | $$ │
│ /$$$$$$$ /$$$$$$ | $$ \__//$$$$$$ /$$$$$$ /$$ /$$ │
│ /$$_____/ |____ $$| $$$$ /$$__ $$|_ $$_/ | $$ | $$ │
│ | $$$$$$ /$$$$$$$| $$_/ | $$$$$$$$ | $$ | $$ | $$ │
│ \____ $$ /$$__ $$| $$ | $$_____/ | $$ /$$| $$ | $$ │
│ /$$$$$$$/| $$$$$$$| $$ | $$$$$$$ | $$$$/| $$$$$$$ │
│ |_______/ \_______/|__/ \_______/ \___/ \____ $$ │
│ /$$ | $$ │
│ | $$$$$$/ │
│ by pyup.io \______/ │
│ │
╞══════════════════════════════════════════════════════════════════════════════╡
│ REPORT │
╞════════════════════════════╤═══════════╤══════════════════════════╤══════════╡
│ package │ installed │ affected │ ID │
╞════════════════════════════╧═══════════╧══════════════════════════╧══════════╡
│ django │ 1.2 │ <1.2.2 │ 25701 │
╞══════════════════════════════════════════════════════════════════════════════╡
│ Cross-site scripting (XSS) vulnerability in Django 1.2.x before 1.2.2 allows │
│ remote attackers to inject arbitrary web script or HTML via a csrfmiddlewar │
│ etoken (aka csrf_token) cookie. │
╘══════════════════════════════════════════════════════════════════════════════╛
--bare
Output vulnerable packages only.
Example
safety review --file report.json --bare
django
What's inside
Installation instructions, usage examples, Docker usage, CI integration, 12 CLI options, and a review subcommand section
Change this for your project
- Replace
safetywith your own package name if you fork the tool - Replace
pyup.ioreferences with your own vulnerability database URL if self-hosting - Replace
requirements.txtwith your actual requirements file name
Where it goes
Load as policy context for the agent, or keep beside the code enforcing the rules.
Worth borrowing
- Using
--stdinto pipe pip freeze or requirement files into the tool - Caching vulnerability database requests locally with
--cache - Outputting JSON for programmatic consumption and later review with
safety review
Related Documents
Guardrails, Safety & Content Filtering
Implements a layered guardrail system with input validation, output validation, and content filtering to protect LLM applications from prompt injection, jailbreaks, and data leaks.
DeepSeek R1: Case Study in Failed Extrinsic Alignment
Compiles public security research and independent findings to argue that extrinsic alignment methods are insufficient for AI safety, using DeepSeek R1 as a case study.
AI Safety & Guardrails for Voice Assistants
Defines a multi-layer safety architecture for voice assistants, covering input filtering, deterministic FAQ routing, RAG-grounded AI responses, and output guardrails.
Risk Assessment Matrix
Documents 12 risks with likelihood, impact, score, level, and mitigation for a child-facing AI app's data protection impact assessment.