Documentation is the backbone of any reliable IT operation, yet it is often the most neglected task because it is time consuming and tedious. Large language models like ChatGPT have changed this dynamic by acting as a highly capable technical writer that understands infrastructure, networking, and security concepts. By providing the model with raw logs, command outputs, or rough notes, you can generate professional incident reports, repeatable Standard Operating Procedures (SOPs), and technical runbooks in a fraction of the time. The key to success lies in how you structure your prompts and how you sanitize your data before processing.
Standardizing Incident Reports from Raw Logs
When a service goes down, your priority is restoration, not prose. However, stakeholders require a clear post-mortem that explains the root cause and the remediation steps taken. ChatGPT excels at transforming messy CLI output into a structured narrative. Instead of writing the report from scratch, you can feed the model your shell history and relevant log snippets.
For example, if you were troubleshooting a DNS issue or a firewall misconfiguration, you might provide the model with the output of your troubleshooting commands. If you are working on network security, you might be comparing different firewall platforms like we discuss in our guide on OPNsense vs pfSense, and ChatGPT can help document why a specific rule was implemented during an outage.
Use a prompt like this to generate a report:
Act as a Senior Systems Administrator. I will provide a series of log entries and the commands I ran to fix a server outage. Generate a formal Incident Report including:
1. Executive Summary
2. Timeline of Events
3. Root Cause Analysis
4. Resolution Steps
5. Recommendations for Future Prevention
Logs: [Paste logs here]
Commands: [Paste history here]This approach ensures that the technical details are accurate while the formatting remains consistent across your entire organization.
Creating Reproducible SOPs for Complex Deployments
Standard Operating Procedures (SOPs) are often out of date the moment they are written. ChatGPT can help maintain these by converting a list of manual steps into a structured document. If you have just finished a project, such as implementing a robust backup strategy following the 3-2-1 backup rule, you can ask ChatGPT to formalize the process for the rest of your team.
The model is particularly good at adding context to commands. If you provide a script, it can break down what each flag does, which is essential for junior staff who might be following the SOP later. When building these documents, always specify the target audience. An SOP for a fellow senior engineer should look very different from a guide intended for a helpdesk intern. You should also ask the model to include a 'Prerequisites' section and a 'Verification' section to ensure the task was completed correctly.
Generating Technical Runbooks with Markdown
Runbooks are more specialized than SOPs, focusing on specific procedures for system recovery or maintenance. ChatGPT can generate these in Markdown format, making them easy to host in GitHub, GitLab, or a local Wiki. Because the model understands syntax, it can write valid configuration files for tools like Ansible, Terraform, or Docker as part of the runbook.
If you are documenting a VPN deployment, you can use the model to create a step-by-step runbook that includes the necessary configuration blocks. For instance, if you followed our WireGuard VPN setup guide, you could ask ChatGPT to generate a runbook for adding new clients to that specific environment. Here is an example of a prompt to generate a technical runbook snippet:
Create a Markdown runbook for rotating SSH keys across a fleet of Ubuntu servers.
Include the specific ssh-keygen commands, the path to the authorized_keys file,
and a bash loop to distribute the keys.
Ensure the tone is technical and concise.This allows you to build a library of scripts and procedures that are formatted identically, reducing the cognitive load on your team during high-pressure situations.
Sanitization and Security Best Practices
The biggest risk when using ChatGPT for IT documentation is the accidental disclosure of sensitive information. You must never paste raw data that contains public IP addresses, cleartext passwords, API keys, or proprietary customer data into the prompt. Before sending data to the model, use a local script or a simple find and replace to anonymize the content.
- Replace real IP addresses with placeholders like 192.168.1.x or 10.0.0.x.
- Change specific server hostnames to generic labels like web-server-01 or db-prod-01.
- Remove any personally identifiable information (PII) belonging to users or clients.
- Use 'dummy' credentials in any code examples or configuration blocks generated by the model.
By following these sanitization steps, you can leverage the power of the model without compromising the security of your infrastructure. Treat the output as a draft that requires a final technical review by a human to ensure that the logic is sound and no hallucinations have introduced dangerous commands.
Want to go deeper?
Our 50 AI Prompts for IT Professionals contains 50 tested prompts for real IT workflows: incident reports, runbooks, client communication, troubleshooting, and change management. $9, instant download.