What is the MCP Hello Page proposal?

A blog post (134pt HN, May 2026) proposed that every MCP server expose a /.well-known/mcp endpoint returning a human-readable and machine-readable description of what the server does, its tools, and how to connect. No MCP spec requirement for this exists yet.

How do agents currently discover what an MCP server can do?

Agents load the full tool schema on connection, which costs thousands of tokens per server. Developers must read source code or documentation. There is no standard lightweight discovery endpoint.

Is there an MCP server directory or registry?

Yes, PulseMCP and the official MCP Registry list 10k-17k+ servers. But these are human-browsable directories, not machine-queryable capability APIs.

What would an ideal MCP server capability page return?

A short structured document describing: server name and purpose, list of tools with one-line descriptions, authentication requirements, rate limits, and a contact or repo URL. Under 500 tokens total so agents can load it cheaply.

Is MCP discovery part of the 2026 roadmap?

The MCP 2026 roadmap mentions capability negotiation and lazy schema loading, but not a standardized discovery endpoint. The Hello Page proposal came from the community, not the Anthropic MCP team.

← Back to dashboard

clawsmith.com/signal/mcp-server-no-human-readable-capability-page-agent-discovery

⚠ IssueWide OpenLive

MCP servers have no standard human-readable or agent-discoverable capability page, so connecting a new server is trial and error

A 134-point HN post (May 2026) proposed an 'MCP Hello Page' standard: a simple /.well-known/mcp page where any server describes itself in plain language. Currently, discovering what tools a server exposes requires reading source code or loading the full schema into an LLM. Agents waste tokens fetching schemas they might not use, and humans have no quick reference for evaluating a server before connecting it.

Product Idea from this Signal

A web app that generates, validates, and hosts machine-readable capability pages for MCP servers so agents and developers can discover what any server does without loading its full schema

197 ▲

Every MCP server today is a black box until you connect to it: discovering what tools it exposes requires reading source code or feeding the full schema to an LLM. The MCP spec roadmap includes Server Cards served at /.well-known/mcp, but no tooling generates, validates, or hosts them. This is a web app where a developer points at their MCP server (or pastes a schema), gets a structured capability card auto-generated, validates it against the draft spec, and receives a hosted page with a stable URL that registries and agents can query without a live connection.

MCPAI AGENTSDEVELOPER TOOLSDISCOVERYSPEC TOOLINGAPI

Competitive40 leadsView Opportunity →

Score Breakdown

197

Social Proof 2 sources

MCP Hello Page

Dachande663 · 5/16/2026

183 HN

Show HN: MCPfinder - An MCP server that finds and installs other MCP servers

coderai · 4/24/2026

Gap Assessment

Wide OpenNo dedicated solution exists

No adopted standard for MCP server capability discovery pages. The MCP spec does not mandate a discovery endpoint. No tooling generates or validates them.

Frequently Asked Questions

Virality Score

197

across 0 platforms

Details

Signalissue

Ecosystem—

Sources2

Platforms0

Updated4h ago

Trend→ stable

Top ideas

All ideas →

0A mobile app that saves articles fully offline with a one-time purchase and no cloud subscription 0A CLI tool that benchmarks AI coding agents against a team's own real production tasks and codebase 0An MCP server that gives AI agents a production-grade browser session with built-in anti-bot bypass, CAPTCHA resolution, and stateful session recovery

Related signals

All signals →

27.9KAI agent tool outputs bloat context window and burn tokens with no compression layer 6KNo stable cross-platform MCP protocol for AI agent mobile automation on iOS and Android 5.7KNo universal standard for agents to share codebase context across tools (AGENTS.md gap)