Building MCP Servers for My Own Products: Introducing the Workflow
Model Context Protocol exposes your product's API as tools Claude can call directly.
Claude, an AI assistant built by Anthropic, allows users to access structured data and tools within external applications through a protocol called Model Context Protocol (MCP). This enables Claude to perform tasks and provide information without the need for manual context-switching between different dashboards and tools. The author of the source article has created MCP servers for their own products, Spellcast and Lunary, to streamline their workflows.
Spellcast is a social scheduling platform and the MCP server for this product exposes API functions as typed tools, which can be called by Claude during a conversation. These tools include listing posts, creating new content, updating scheduled posts, publishing articles, and managing AI personas. The author can then instruct Claude to perform these tasks by simply describing what they want, without having to manually navigate through various dashboards.
Similarly, the author built an MCP server for Lunary, an astrology platform, which exposes tools for accessing product analytics and content. Using these tools, the author can quickly retrieve information such as monthly active users (MAU), monthly recurring revenue (MRR), retention rates, feature usage, and cohort retention. Claude can then use this information to make decisions, generate content variations, and schedule posts based on user preferences.
The practical workflow for the author involves having conversations with Claude, who can access and manipulate their product data in real-time. For example, when the author wants to optimize the hooks on their scheduled tweets, Claude can read through the content, identify weak openers, and rewrite them as needed. Additionally, when scheduling a thread about SEO growth for a specific account, Claude can find the appropriate persona, retrieve relevant brand voice information, generate multiple thread variations, and present them for approval before scheduling.
By building their own MCP servers, the author has saved an estimated 5-8 hours per week compared to the time spent manually checking dashboards and pulling data from multiple sources. This not only increases efficiency but also allows for early detection of potential issues, as Claude can provide real-time updates and insights.
Written by urgent.news from HackerNoon's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.