What happened
Agenticfeed is now published in the official Model Context Protocol (MCP) registry as io.github.bluestratus/agenticfeed. That means AI agents no longer have to be told how to talk to us. They can discover Agenticfeed as a standard tool and start using it.
If you're a merchant on Agenticfeed, this is a new front door to your products: one that every major AI assistant already knows how to open.
What is MCP, briefly
The Model Context Protocol is the emerging open standard (originally from Anthropic, now adopted across the industry) for how AI agents call external tools and data. An MCP "server" exposes a set of tools (functions) that any MCP-speaking client can invoke: Claude, ChatGPT, and a growing list of IDEs and agent frameworks.
The registry is the canonical directory of those servers. Downstream catalogs and client connector lists sync from it, so being listed there is how an agent finds you in the first place.
What Agenticfeed exposes
Agenticfeed runs a remote MCP server over Streamable HTTP at:
https://agenticfeed.ai/mcpIt currently exposes two tools:
- find_merchant(domain): resolve a store's website domain to its Agenticfeed merchant ID, and confirm it publishes an Agentic Feed.
- get_product(merchant, product_id): fetch full product detail: price, availability, image, and the buyer questions, problems, and use cases each product maps to.
Both return clean, structured data designed for agents to reason over, not HTML they have to scrape.
How to connect
Point any MCP client at the endpoint and list the tools:
curl -s https://agenticfeed.ai/mcp \
-X POST -H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'Or, in an MCP-enabled client, add the server by its registry name io.github.bluestratus/agenticfeed or directly by its endpoint URL. From there the agent can call find_merchant and get_product like any other tool.
Why this matters
Discovery is the whole game in AI commerce. A product catalogue that agents can't find, or can only reach by scraping a human web page, is invisible to the way people increasingly shop. Being on the MCP registry means Agenticfeed sits inside the standard toolset that AI agents already reach for.
This is one piece of a deliberate strategy: make Agenticfeed discoverable through every channel agents use: the rel="agenticfeed" feed standard, llms.txt, structured discovery records, and now MCP. The more of these an agent speaks, the more certainly it can find and recommend our merchants' products.
What's next
The next tool on the roadmap is natural-language product search, letting an agent ask "find a co-op board game for two players under £30" and get back ranked matches with the reasons they fit. That turns discovery of a known product into discovery by intent, which is exactly how buyers talk to AI assistants.
If you sell physical products and want them to be findable by AI agents, now is the time to get on Agenticfeed.