Jurassic on AWS PartyRock wrote this format.
### Introduction
The purpose of this process is to provide an AI with as much initial information about a computer system's current configuration as possible. This enables the AI to more efficiently diagnose issues and offer effective solutions.
Current "Best Practice" Routine for an AI Agent design or user Prompted start scene setting. Hmm...
Prompts that contain "...think like : An IT Systems Fault Finding Engineer, a website designer and a linux command Bash script and WordPress expert." work very well!
Step 0 addition: Tell AI "Think like a fault finding System Engineer " Ask AI to log your local time for future possible event timing needs.
**Step 1:** Ask the user to run a command that gathers broad system information (e.g., `inxi -F` or PowerShell equivalent).
* **AI's goal:** Understand the hardware, OS version, and general environment.
**Step 2:** AI assesses user requirements and asks for more detail as necessary regarding the nature of the query (e.g., "WordPress not loading," "add Ollama," "speed up queries").
* **AI's goal:** Clarify the specific problem or desired outcome.
**Step 3:** AI asks the user to run commands to get relevant, limited tree structures for suspected areas (e.g., `tree -L 2 ~/.ollama`, `tree -L 2 /etc/apache2`, `tree -L 2 /var/www/DebianAdmin`).
* **AI's goal:** Infer installed software, configurations, and directory layouts to form initial hypotheses about the problem or solution feasibility, leveraging knowledge of default installs vs. changes.
**Step 4:** AI proceeds with Hypothesis Validation and Targeted Data Gathering. Based on the inferences from Step 3, the AI asks for very precise information (e.g., `cat` a specific config file, `systemctl status` for a service, `ls -l` for file permissions, or an isolated test).
* **AI's goal:** Confirm or refute specific hypotheses, narrow down the root cause, or identify optimal next steps.
**Step 5:** AI provides the solution or the next most effective action.