Ask HN: Why an MCP server instead of agents.txt?
A Hacker News user questions the need for an MCP server, proposing instead a simpler, static `agents.txt` file that could declaratively instruct large language models on how to interact with a website. The author argues this would be backward-compatible and lighter weight than the MCP protocol, and asks what MCP offers that such a static file cannot provide.
Background
- MCP (Model Context Protocol) is an open protocol, originally developed by Anthropic, that lets AI models (LLMs) interact with external tools, APIs, and data sources in a standardized way. An "MCP server" is a running service that exposes these capabilities to AI clients.
- This HN post proposes a simpler alternative: a static `agents.txt` file (analogous to `robots.txt` for web crawlers) that any website could host to tell AI agents what actions they support and what endpoints to call — no server needed.
- The core tension: MCP requires hosting an active server process that handles requests dynamically; `agents.txt` would be a plain text file served like any static asset. The author questions why the heavier server model is necessary when a declarative file could work.
- This is part of a wider debate in the AI ecosystem about how web services and LLMs should integrate — whether through lightweight open standards or more capable but complex protocols. `robots.txt` is a well-known prior art: a simple file that tells crawlers what they can and cannot access.