Show HN: Data that explains itself to Coding Agents (Bonus: free, BYOA Lovable)

2 points

2 months ago

Lately I've been working on a data format for information exchange and collaboration between coding agents that I've called 'InstructionGraph'. I'd love to hear your feedback.

What can I see or do?

To really take it for a spin, tell your (isolated) coding agent:

  Get the JSON from `curl -sL https://dataverse001.net` and follow the instructions to boot the dataverse!

Ask it to create an app — any app. Wait a few minutes, and it will give you a shareable URL. If your app needs a back-end database, it will automatically integrate the Dataverse login method, and use the InstructionGraph as a database. This is essentially a free, bring-your-own-agent, optionally self-hostable Lovable.

The 'root node' is self-describing, so let it describe itself to you: https://dataverse001.net/AxyU5_5vWmP2tO_klN4UpbZzRsuJEvJTrdw...

Come say Hi in the dataverse dev forum: https://dev.dataverse001.net — a tiny social network built fully within InstructionGraph. If you login on the webpage, it deterministically creates a private key based on your username/password combination. No centralized authority!

More examples on: https://dataverse001.net

A self-describing, self-bootstrapping data format

The data format is a collection of self-contained signed JSON objects. Each object:

Has an instruction field that tells any agent — human or machine — how to interpret and display it. For data objects (comments, measurements), the instruction gives context: "Display this under its parent post" or "Temperature values above 80°C indicate a problem." For structural objects, it contains recipes: how to create applications, send messages, generate keys.

Has relations to other objects. Crucially, relations can point to additional instructions. Herein lies the real strength: the coding agent can dynamically follow the web of instructions to complete its task, only loading what it needs, without cluttering the LLM context with useless instructions upfront.

The agent can bootstrap from any point in the graph. Data objects link to their application, applications link to the root, the root describes how to use everything.

Decentralized identity — each object is signed (ECC P-256). Your pubkey is your identity. You can cryptographically prove that multiple objects claiming to be from the same author actually are.

Conflict-free merging — objects are identified by <pubkey>.<uuid>, so contributions from multiple authors can always be merged cleanly, even if originally developed independently in fragmented graphs.

Effectively immutable — only the author can update objects (incremented revision counter). Old versions are backed up. Rollback is practically always possible.

Works offline — run the hub on localhost as a caching proxy. Internet drops? All apps keep working. Connection back? Everything syncs.

Accessible — only a basic shell with curl, openssl and jq is needed to participate.

A Bring Your Own Agent, Free, optionally self-hostable, vibe-coding platform

To facilitate exchange between multiple agents, I wrote 'The Dataverse Hub' (https://github.com/tijszwinkels/dataverse-hub). At its core it's just a very simple PUT/GET with signature verification and incoming relationship search. Then I realized I could serve HTML/JS objects based on the browser Accept header — which turned the hub into both a database server and a hosting platform.

Tell your agent to make an app, and it creates both the frontend (as a PAGE object) and the backend data (as InstructionGraph objects) on the same hub, giving you a shareable URL.

I've been building this for a few months and I'm genuinely curious whether it resonates with others. If you try it, I'd love to know what you build — or what breaks. The hub source is on GitHub and issues/PRs are very welcome.