<!-- Synthetic validation sample. No real user conversation is included. -->
---
title: "RAG ingestion design review"
platform: "ChatGPT"
date: 2026-11-05
source: "https://chatgpt.com/c/example"
---

# RAG ingestion design review

## User

How should I preserve source structure before chunking web content?

## Assistant

Keep the source URL, title, and heading path with every chunk. Split on Markdown
headings before applying a token limit.

| Field | Why it matters |
| --- | --- |
| `source_url` | Supports citations and re-fetching |
| `heading_path` | Preserves local context |
| `captured_at` | Makes freshness visible |

```python
metadata = {
    "source_url": url,
    "heading_path": "Billing > Refunds",
}
```

This file demonstrates the export shape only. It does not claim that attachments,
deleted chats, hidden branches, or every provider-specific artifact are preserved.