Publishing in an MCP Era

Tinkering with Cloudflare’s MCP stack (thanks, Carlos!) makes me wonder: What would an MCP-native publisher look like?


Test mine out here; it’s full of information about me.

By ‘publisher,’ I mean anyone: bloggers, experts, even journalists.

Web search is collapsing, and social link-outs were killed a while ago (except on platforms like Farcaster).

Publishing MCP-native content would mean formatting, tagging, and designing for machines reading, not human scanning.

The main challenge is distribution. Most consumer LLMs still require users to pull information rather than pushing it proactively. I expect this’ll shift with enough time though; the workflows and tools are all being built in enterprise spaces already.

(Side note: let’s assume using MCPs will get significantly easier, rather than the awkward copy-pasting of JSON it involves right now.)

Then how do you stay relevant and convey your voice if LLMs are your only path to the user? I suspect we’ll see style or usage guides embedded into MCPs—maybe even with XML-like tags. Beyond that, I think it’s about being prolific: writing a high volume of content consistently will increase the surface area for LLMs to seek out and connect.

That is: the more prolific, niche-aligned, and well-formatted your content, the more agents will grab you.

And the final big question is how would you monetize? If MCPs adoption evolves with a recommendation system, in theory organic customer acquisition cost could be very low. You’d likely want to boost that through some additional marketing spend. This feels similar to YouTube, where the sponsor and / patron model is seemingly sustainable for a wide range of creators. Alternatively, publishers could embed sponsorship and calls-to-action directly in an MCP module, ready to be surfaced in LLM responses.

There may be short-term attention arbitrage, especially for technical audiences, for publishers of any kind to experiment with MCP-native publishing.* *You’d at least learn a bunch along the way about consumer attention and LLM distribution.