Naive API-to-MCP conversion

Organizations are eager to let AI agents interact with existing systems, often by attempting a seamless, direct conversion of internal APIs to the Model Context Protocol (MCP). A growing number of tools, such as MCP link and FastAPI-MCP, aim to support this conversion.

We advise against this naive API-to-MCP conversion. APIs are typically designed for human developers and often consist of granular, atomic actions that, when chained together by an AI, can lead to excessive token usage, context pollution, and poor agent performance. In addition, these APIs — especially internal ones — frequently expose sensitive data or allow destructive operations. For human developers, such risks are mitigated through architecture patterns and code reviews, but when APIs are naively exposed to agents via MCP, there’s no reliable, deterministic way to prevent an autonomous AI agent from misusing such endpoints. We recommend architecting a dedicated, secure MCP server specifically tailored for agentic workflows, built on top of your existing APIs.