KaggleIngest User Guide - How to get rich context from Kaggle competitions for AI coding assistants (Claude, ChatGPT, Cursor, Copilot). Step-by-step guide with output format comparison.

kaggleingest_user_guide.md

How to Use KaggleIngest

Get rich context from Kaggle competitions for your AI coding assistant in 3 simple steps.

Website: https://kaggleingest.com

---

Step 1: Get Your Kaggle URL

Copy the URL of any Kaggle competition or dataset you want to analyze:

https://www.kaggle.com/competitions/titanic
https://www.kaggle.com/datasets/uciml/iris

---

Step 2: Paste and Configure

1. Go to kaggleingest.com
2. Paste your Kaggle URL in the input field
3. Configure options:
   • Top N Notebooks: How many top notebooks to include (default: 10)
   • Output Format: Choose toon (token-optimized), json, or txt
   • Dry Run: Preview what will be extracted without downloading

---

Step 3: Download Your Context File

Click "Get Context" and wait for processing. Once complete:

1. Click "Download" to get your context file
2. Open the file and copy its contents
3. Paste into your AI assistant (Claude, ChatGPT, Cursor, Copilot, etc.)

Your AI assistant now has:

• Competition metadata and evaluation criteria
• Top notebook insights ranked by quality
• Dataset schemas and column descriptions
• Key code patterns and approaches

---

Output Formats

Format	Best For	Token Usage
toon	LLMs (recommended)	~40% fewer tokens
json	Programmatic use	Standard JSON
txt	Human reading	Plain text

---

Tips

• Start with Titanic: Try https://www.kaggle.com/competitions/titanic to test
• Adjust Top N: For large competitions, start with 5-10 notebooks
• Use TOON format: Saves context window for your LLM's responses

---

Need Help?

• Discord: https://discord.gg/EYTXJC56a
• GitHub: https://github.com/Anand-0037/KaggleIngest/issues
• Blog: https://dev.to/anandvashishtha/building-kaggleingest-how-i-bridged-kaggle-data-with-ai-coding-assistants-1j8b