Re: ADP Update (POC WIP)

update; https://webcivics.github.io/Webcard/?domain=sailingdigital.com

On Tue, 8 Jul 2025 at 23:29, Timothy Holborn <timothy.holborn@gmail.com>
wrote:

> Hi All,
>
> I've made some progress on 'Agent Discovery Protocol' (ADP) whilst still a
> Work in Progress (WIP); here's a link to the Proof of Concept (POC) demo
> app.
>
> BACKGROUND: My Presentation at IETF:
>
> https://www.ietf.org/proceedings/119/slides/slides-119-hotrfc-sessa-09-agent-discovery-protocol-00.pdf
> https://www.youtube.com/watch?v=ZDH5eaIF_zc&t=2276s
>
> UPDATE & DEMO:
>
> https://webcivics.github.io/IPFS-ADP/
> (FWIW: you can use sailingdigital.com to check if you'd prefer not to
> produce your own DNS record, to see the output)
>
> In this situation, I've used IPFS to store the file.  There's lots that
> can be done to make it nicer.  but for now, it works well enough to
> illustrate a few points.
>
> 1. Agent Types
> - natural Person - this is intended to provide an anchor, which is
> in-part, intended to be interoperable with WebID (or multiple WebIDs) and
> in-turn also, enable the means to present a POD location (ie:
> mypodprovider.tld/mypodaddress) - which might change!  theory being, if
> they retain access to their domain, then address book apps, etc. can
> rediscover any changes.  Another intended purpose of it, is to provide
> bitcoin addresses (and related hash's) so that apps can create
> functionality where people just share their own domain name; and the
> address can be found via ADP.
>
> NOTE: This is not intended to store private information.  The intended
> goal, is to ensure that people can own a domain name; but then, provide
> information about their WebID / POD (or however their 'mydata' or related
> solution is made to work) address - where connections can be made, and the
> permissive information shared between peers (on an agreed basis, etc.).
> This is also intended to be made AS SIMPLE AS POSSIBLE.  Ideally, users
> only need a domain name and means to add a DNS record - rather than
> otherwise REQUIRING a hosting account.
>
> - organisations - fairly straight-forward...
>
> - AI agent - my thinking is that this is useful in particular for AI
> generated Content, if it retains some information about where its from;
> whether it be, text, image, etc.  this could in-turn create a means to
> discover, for instance, what AI Agent a piece of content is from.
>
> - Humanitarian Service: The point of this was to address internet shutdown
> related issues.  So, rather than shutting the internet down entirely -
> perhaps there's a means to ensure that at least critical online resources
> remain accessible.
>
> - Adult Content Website; point is - a means for websites to declare
> themselves an 'adult website' in a way that then enables settings on local
> machines to prevent access to those sites (ie: preventing children from
> accessing adult sites or perhaps even being able to find them in
> search-results).
>
> 2. Defining Properties (in RDF); this also requires some consideration
> about ontology...
>
> 3. Web of Trust:  old records -
> https://www.w3.org/wiki/index.php?search=%22web+of+trust%22&title=Special%3ASearch&profile=default&fulltext=1
>
>
> The notion of Web of Trust is that if peers (ie: different domain names)
> declaratively state something about one-another (forming a graph) then
> there's a means to improve assumptions.  This might be used, for instance,
> by a series of institutional websites all denoting that a particular
> humanitarian domain is in-fact a humanitarian (IHL) related domain name...
>
> This part needs more work.  I'm still thinking about how it might better
> be applied.  Earlier work considered the use of credentials, which is still
> assumed to be part of what ADP should support - but maybe, 'web of trust'
> related methods provides a decentralised alternative - there's various
> implications, as noted, still thinking about it.
>
> Generation & Implementation:
> I like turtle.  I know, most platforms prefer JSON-LD or moreover JSON -
> but, maybe that's making it more difficult to distinguish between -
> ideologies?  idk.  For the most-part, serialization formats are
> interchangable.  anyhow,
>
> app generates the file, you then upload it to IPFS, get the CID generated
> when uploading the file, then enter into the field provided.  The app then
> generates the DNS records needed to be added (to DNS).
>
> I found it worked after - less than a minute, although the update took a
> little longer.
>
> I'm particularly interested in feedback, particularly technical feedback -
> I'm presently unsure about security related considerations in particular,
> whilst also considerate of the importance of better defining the ontology
> (indeed also, namespace for it)..
>
> It would also be good to get some feedback about whether this is something
> that the CG participants want to collaboratively pursue or not.  As has
> been noted earlier, I think (suspect?) part of it probably needs to be done
> also via IETF.  I'm still working on the ID (Internet Draft).
>
> 🙏🕊️
>
> Timothy Charles Holborn.
> Media Prophet | Web Civics | Trust Factory | Webizen
> https://www.linkedin.com/in/ubiquitous/
>