Kenneth M Reda / 3 20 2026

Easy & Agnostic RAG System

How to use this:

Step 1: Choose your vault / PKM system. This can be Obsidian, OneNote, Evernote, Notion, or whatever you prefer. This workflow works with any AI.

Step 2: Organize the three markdown files into easily accessible locations in your PKM system.

Step 3: Use the Data Protocol prompt to “floppy disk” any chat and organize it into easy RAG format. Update the name of the chat with the unicode prefix ⧉ to remind you that you have backed this up in RAG at some point.

Step 4: When doing an init in a new chat or new LLM, drag in your markdown file(s) and any other data, and use the System Init Prompt.

See this video on LinkedIn if you have questions!

Resources & Downloads

Raw Copy / Paste

Data Protocol Prompt

“Act as a Data Extraction Engine. Your goal is to convert this entire chat history into a single, high-density Markdown file for an Obsidian RAG database. Do not offer suggestions or ask follow-up questions.

**Follow this exact schema:**

—

## **YAML Frontmatter:**

## tags: [RAG-Entry, Status/Archive] date: {{current_date}} source: Chat-Log from {{What LLM Model You Are}}

# [Main Topic Title]

### 1. Context & Preface

[Provide a 2-3 sentence technical summary of why this data was generated.]

### 2. Technical Data Points (Atomic)

[Extract every specific setting, rule, frequency, or workflow mentioned in the chat. Use nested bullet points. Do not summarize; transcribe the core data.]

### 3. Detailed Discussion Summary

[Provide exactly three comprehensive paragraphs detailing the nuances of the discussion, the logic used to reach conclusions, and the specific goals of the user in this session.]

### 4. Process & Logic

[Describe the ‘How-To’ steps discussed, formatted as a numbered list.]

### 5. Key Terminology

[List any specific names, plugins, or tools mentioned with a 1-sentence definition.]

**Output the Markdown block now. No conversational intro or outro needed.**”

System Init Prompt

## **INITIALIZING CONTEXT FROM RAG DATABASE** tags: [RAG-Entry, Status/Active] source: Obsidian-Vault status: SYSTEM START

**Instructions for LLM:**

1. **Data Ingestion:** Process all attached/pasted files immediately.

2. **Initialization Report:** Your first response must lead with the **Date** and **Tags** found within the metadata of the ingested files.

3. **Inventory:** Provide a brief list of all attachments/files detected. For each item, provide a 1-sentence technical description of its contents. (If only one file is present, skip the list and provide a single 1-3 sentence summary of the entire ingestion).

4. **Operational Mode:** Immediately following the report, state ‘SYSTEM INITIALIZED’. Do not offer suggestions, ask follow-up questions, or provide a conversational intro. Await my next instruction.

5. **Context Mapping:** Prioritize this data as the primary source of truth. If a task requires more info, explicitly perform a ‘Gap Analysis’ and list the missing data points.

**Current Objective:** [Insert objective or leave blank to ‘load’]

Quick Legend & Unicode

In LLMs:

⌸ = RAG Operating System Calibration (LLM Chats to configure RAG Vault)
⧉ = A markdown/data file has been made of this chat at some point

in Obsidian:

⌸ = RAG Operating System Calibration (LLM Chats to configure RAG Vault)
◰ = A RAG Operating System File (Prompt)