Why not ask the agent to read .well-known/agent.md (or better yet something.com/agent.md) and do a bootstrap if it is allowed! to do so. It can download skills, configure mcps, etc.
The best solution we could find for this problem was a combination of a Wiki + "Saved Programs" (executable scripts).
When you ask a question, the system does a fuzzy Wiki search to see if any topic has relevant info. Wiki links serve as "graph edges" that form a Knowledge Graph.
So lets say I ask "What products from the FOOBAR API sold least last month?"
The agent would look up "FOOBAR API", and then write a script to call the "products" endpoint with a date range + SQL fetch from "sales" DB and do whatever it needs to do.
If none of this info exists (URL to FOOBAR, location of "sales" data), the agent asks for more info and offers to update the wiki for future.
Very close to this at my company. We have docs each of which having its description in the yaml. An MCP tool lists the doc sections, a full section with its descriptions, or the entire contents of a doc. A kind of progressive disclosure. Works really well. We even write "skills" this way so they can be used in all our chat environments.
It feels like there's a lot of overlap with the existing MCP "Resources" [0] concept, and that "Searchable Registry" + multi-protocol (MCP/A2A, etc) is the main difference?
Too many protocols got created (MCP, A2A, etc) and so now you've got to create a new standard that can consume them all, and Google being a search index would like to index them, please =)
This is search. Please don't reinvent search with a new acronym.
Agents can use Google. They can also see all the same signals humans do to judge quality. Maybe there's a need for a specialized directory, but does it need a spec or an acronym? No.
This seems useful as a standard interface, but, again, doesn’t solve the harder discovery problem. Once there is value in appearing in the top results, the system inherits the same adversarial dynamics as SEO/app stores, spam, scams, etc.
Why not ask the agent to read .well-known/agent.md (or better yet something.com/agent.md) and do a bootstrap if it is allowed! to do so. It can download skills, configure mcps, etc.
There is no need for another file IMHO.
I work at a company with an LLM-adjacent tool.
The best solution we could find for this problem was a combination of a Wiki + "Saved Programs" (executable scripts).
When you ask a question, the system does a fuzzy Wiki search to see if any topic has relevant info. Wiki links serve as "graph edges" that form a Knowledge Graph.
So lets say I ask "What products from the FOOBAR API sold least last month?"
The agent would look up "FOOBAR API", and then write a script to call the "products" endpoint with a date range + SQL fetch from "sales" DB and do whatever it needs to do.
If none of this info exists (URL to FOOBAR, location of "sales" data), the agent asks for more info and offers to update the wiki for future.
Very close to this at my company. We have docs each of which having its description in the yaml. An MCP tool lists the doc sections, a full section with its descriptions, or the entire contents of a doc. A kind of progressive disclosure. Works really well. We even write "skills" this way so they can be used in all our chat environments.
It feels like there's a lot of overlap with the existing MCP "Resources" [0] concept, and that "Searchable Registry" + multi-protocol (MCP/A2A, etc) is the main difference?
Too many protocols got created (MCP, A2A, etc) and so now you've got to create a new standard that can consume them all, and Google being a search index would like to index them, please =)
[0]: https://modelcontextprotocol.io/specification/2025-11-25/ser...
As always, relevant XKCD https://xkcd.com/927/
This is search. Please don't reinvent search with a new acronym.
Agents can use Google. They can also see all the same signals humans do to judge quality. Maybe there's a need for a specialized directory, but does it need a spec or an acronym? No.
MCPs: Json endpoints with a "description" field.
More like: JSON-RPC, but without all good stuff.
But "search" doesn't have "agentic" in it. But oh wait, we can make it:
Super Explanatory Agentic Resource Catalog Heurestics - SEARCH
This seems useful as a standard interface, but, again, doesn’t solve the harder discovery problem. Once there is value in appearing in the top results, the system inherits the same adversarial dynamics as SEO/app stores, spam, scams, etc.