If you are using a Debian-like distro and are a new user or just starting your career as a sysadmin, you probably already know the importance of updating packages using apt update. You might also want to understand what each package does to learn more about Linux. Additionally, system administrators (sysadmins) often need to communicate or document to stakeholders which updates are urgent or security-related.
In this post, I will show you how to combine Python, the apt list -u command, and Gemini AI to create human-readable summaries of pending package updates.
apt list -u | awk '{ print }' | sed 's|/.*||'
Here’s a breakdown of the solution across two scripts:
This script runs apt list -u to fetch pending updates, processes the output, and uses the prompt function to get categorized summaries from Gemini AI.
import subprocess from utils.gemini_cfg import prompt try: # Run 'apt list -u' to list upgradable packages result = subprocess.run(["apt", "list", "-u"], capture_output=True, text=True, check=True) output = result.stdout # Get command output # Use the Gemini AI model to summarize the updates summary = prompt(output) # Save the AI-generated summary to a Markdown file with open("./gemini_result.md", "w") as file: file.write(summary) print("Summary saved to gemini_result.md") except subprocess.CalledProcessError as e: print("Error while running apt list:", e)
This script configures the Gemini API and defines the prompt function for AI-generated content.
import google.generativeai as genai from environs import Env # Load API key from .env file env = Env() env.read_env() key = env("TOKEN") # Replace with your environment variable key name # Configure Gemini API genai.configure(api_key=key) model = genai.GenerativeModel("gemini-1.5-flash") # Function to prompt Gemini AI for summaries def prompt(content): message = ( "You work as a sysadmin (Debian server infrastructure). " "You must create a list categorizing the importance in terms of security and priority, " "providing a brief summary for each package so that business managers can understand " "what each library is from this output of the `apt list -u` command: " f"{content}" ) response = model.generate_content([message]) return response.text
The script does the following:
Open gemini_result.md to see a clear, categorized summary of updates for easy communication.
Here's an example of what the generated summary might look like:
## Debian Package Update List: Priority and Security The list below categorizes the packages available for update, considering their importance in terms of security and business operation priority. The classification is subjective and may vary depending on your company's specific context. **Category 1: High Priority - Critical Security (update immediately)** - **linux-generic, linux-headers-generic:** Critical kernel updates to fix security vulnerabilities. - **libcurl4:** Resolves potential security issues for data transfer operations. ... **Category 2: High Priority - Maintenance and Stability (update soon)** * **`e2fsprogs`, `logsave`:** Packages related to ext2/ext3/ext4 file systems. Update to ensure data integrity and file system stability. **Medium-High priority.** ... **Category 3: Medium Priority - Applications (update as needed)** * **`code`:** Visual Studio Code editor. Update for new features and bug fixes, but not critical for system security. * **`firefox`, `firefox-locale-en`, `firefox-locale-pt`:** Firefox browser. Updates for security fixes and new functionalities. Priority depends on Firefox usage in your infrastructure. ...
With a bit of Python and Gemini AI, you can automate and improve how you communicate Debian package updates. This script is an excellent foundation for integrating AI into sysadmin workflows. This post is for educational purposes, so be mindful of Gemini API resources, as well as the secure handling of your system.
Thanks for reading! ?
The above is the detailed content of Automating Debian Package Update Summaries with Python and Gemini (gemini--flash). For more information, please follow other related articles on the PHP Chinese website!