Flatten JSON to a hierarchical EAV table, explore, edit, and convert back to JSON!

Hey Coda!

Just built a system that lets you flatten ANY JSON into a hierarchical EAV table and then convert it back to JSON after editing - all within Coda! Check out the demo video :slight_smile:

Why I built this

Working with complex JSON in Coda has always been challenging. If you’re working with:

  • API responses
  • LLM outputs/inputs (my primary use case)
  • Configuration files
  • Complex data structures

You know the pain of navigating nested JSON structures. You either have to flatten everything into CSV-like tables (losing the hierarchy) or work with JSON as raw text.

How it works

  1. Import: Paste any JSON into the input area and click ā€œFlatten to EAVā€
  2. Explore: The JSON gets transformed into a EAV table where:
  • Each key becomes a node/row
  • Parent-child relationships are preserved
  • Values are stored and editable
  • All relationships (ancestors, descendants, siblings) are tracked
  1. Navigate: Use the Miller Columns view (like Mac Finder’s column view) to click through the hierarchy
  2. Edit: Update any values directly in the table
  3. Export: Click ā€œReconstruct JSONā€ to convert your EAV back to properly formatted JSON

Here’s what makes this special:

  • Full hierarchy preservation - even deeply nested structures
  • Edit any value and see changes reflected in the final JSON
  • Visualize complex structures with Miller columns
  • No depth limits to the hierarchy

Technical Notes

  • The system converts arrays to objects with [n] notation as parents. This keeps the original order while making everything consistently navigable.
  • Since it’s built on Coda tables, there are no practical size limitations beyond Coda’s standard constraints.

Get the doc

You can duplicate this document and start using it right away:

Use cases

I’m currently using this to:

  • Debug complex API responses
  • Create structured prompts for LLMs
  • Store and edit hierarchical data that needs to be accessed by both humans and AI

Would love to hear how YOU might use this!

Happy coding!
Jon

7 Likes

Well done!
This will be very useful to explore json responses, thank you :clap:

2 Likes

Great work, thanks for sharing :folded_hands: :clap:

1 Like

Nice work!

Personally I would either use some robust JSON tree viewer app (for trees over 10-ish kilobytes or complex hierarchies) or a text editor like Notepad++ or VS Code for anything simple :slight_smile: I can imagine it takes quite some time to ā€˜unfold’ that tree in Coda (creating all those rows and stuff — even in your edited loom it takes a while) so for me it’s not super practical. But it’s a great exercise and a novelty doc for sure!

Make it compile schemas for objects of different types though — maybe even some TS code or Schemas for Coda Packs — and now we’re talking

1 Like

Yes, agreed @Paul_Danyliuk! I don’t think the real application of this is exploring/editing nodes in JSON.

For me, the real usecase is being able to update my database with an LLM responding in JSON.

TS/schemas is a good rec! I’ll see what I can do :wink: