Setting Up MCP's for our Agent Builder
MCP’s (Model Context Protocols) are a critical tool needed our out PBS (personal brand site) Agent Builder to work. MCP’s allow LLM’s like ChatGPT, Gemini, Grok etc to interact with external data sources, as connectors. This enhances the abilities that LLM’s have as it allows them to go and gather specific data, like a universal connector. API’s preform a similar role that MCP’s do however, the main difference comes in usability. Before MCP’s existed, for LLM’s to go and gather data from an external source a specific/custom API was needed for every system. For example. For an LLM to go and grab information about you from Youtube, or Google for building personal branding or compmany websites, they each had to have their own API key configured. Now with MCP’s, platforms like Youtube and Google can connect with an LLM via one MCP instead of two, this removes a ton of complexity. So a simple analogy would be that MCP”s add plug and play capabilities to LLM’s for grabbing data. A side not on the name too, Model Context Protocol. These tools have a “protocol” they follow to grab “context” from different Large Language “Models”. If you break it down it totally makes sense.
For our Agentbuilder we have set up mcp’s that go and gather information form Youtube, google and socials to grab information about you, and then we pump that information through a number of agents to fill your website and create blog posts on your site. All of which leads to having more authority on Google and being able to claim your knowledge panel. Inside agent 1 is our PBS discovery agent, we give it our credentials, and then via the MCP we have set up it will go and call those tool to pull data from google and youtube.

You can see here that this MCP we have set up in Agent 01, it will call three different tools that we have configured. These agents that go and fetch this information the out put it in a specific schema. 
This is the code for our Youtube Search tool.

The information gathered then gets sent to Agent 02 to score the authority of it, and then hand off to be turned into content. We have used our GPT’s like Jennifer to configure this data we pull into content that will fill your site.
