Proposal: Start by Defining Design Principles for the Registry Standard

[ognis1205]

How can we structure the AI Registry Standard conversation to avoid chaos?

I'm not a core maintainer of this project, but as someone who has been following the discussions around AI Card and AI Registry topics, I'd like to ask for guidance on how we can approach the upcoming Registry Standard related conversations in a way that avoids the kind of chaotic discussions we've seen in other places (such as previous A2A Registry discussions).

This is not a criticism of anyone, I simply want to contribute in a constructive way, and I'm hoping to learn from the experience of others here.

Orthogonality with AI Card

In general, I see the AI Card specification and the Registry Standard as largely orthogonal. AI Cards act as a standardized envelope for protocol-specific payloads, while the Registry Standard defines discovery and metadata mechanisms.

Because of this orthogonality, I believe it's reasonable for the Registry Standard discussions to progress in parallel, without needing to wait for the AI Card work to be fully finalized.

That said, previous proposals included concepts like Open Discovery, so depending on how the AI Catalog is handled, the boundary may not be completely orthogonal. This makes it even more important to clarify design principles early.

How we might mitigate confusion

One suggestion is to follow the same approach taken in the modelcontextprotocol/registry project, which began with defining clear Design Principles:

https://github.com/modelcontextprotocol/registry/blob/main/docs/design/design-principles.md

Starting with design principles can:

• establish shared expectations early,
• help avoid overly implementation-specific debates,
• and keep the discussion focused and productive rather than chaotic.

I believe this approach would also respect the point that LF projects should remain community-driven and avoid any two-tier structure of "official vs community" efforts.

Next step proposal

I'd like to propose the following next step:

Start a PR that introduces initial Design Principles for the AI Registry.

Not to finalize them, but simply to start the conversation with a shared document that others can review and comment on.

Ultimately, there will be moments where core maintainers will need to exercise judgment and make decisions, but having a principled foundation should reduce friction and avoid the chaotic dynamics we saw in earlier registry discussions.