Build an AI App with the Vercel AI SDK and Strapi MCP

A tutorial explains how to build a Next.js AI content assistant that connects Claude, through the Vercel AI SDK, to Strapi’s MCP server so editors can manage CMS content in a chat interface. It configures Strapi’s MCP endpoint and adds three read-only custom tools: one that finds published articles in the same category, one that checks rich-text Markdown links for missing or unpublished internal targets, and one that uses SerpApi to research Google results for an article topic. The Next.js app keeps the Strapi Admin token server-side, restricts the MCP tools available to the agent, renders custom result cards, and requires explicit human approval before the agent can run update operations, which modify drafts rather than published content. The guide emphasizes registering tools during Strapi’s register phase, sending complete block arrays when updating articles to avoid accidental deletion, validating tool output in the UI, and limiting permissions to read, create, and update capabilities. It also notes that the example lacks authentication and should not be publicly deployed until login, per-editor access controls, rate limits, guardrails, telemetry, and testing are added.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.