Skip to content

ntfy-mcpNotifications your MCP client can send, read back and correct

An MCP server for ntfy: publish notifications, read back what was sent, revise one in place while a job runs, and administer accounts and topic access.

ntfy-mcp architectureAn MCP client talks to ntfy-mcp over stdio; ntfy-mcp publishes notifications and polls the message cache over HTTPS against an ntfy server, which pushes them to the subscribed devices. Because a published message keeps its sequence id, the same notification can be revised in place instead of sending another one.MCP clientClaude, Codex, …ntfy-mcp13 toolstopic allow-listasks a personntfypub-sub serverstdioHTTPS · publishHTTPS · pollsubscribed devices get the pushupdate_message revises it in place
One stdio server between the MCP client and ntfy — publishes over HTTPS and polls the message cache over the same API, bounded by the topic allow-list, with destructive calls put to a person first.

See it in action ​

Demo of ntfy-mcp over the MCP inspector

Running it elsewhere ​

A client that cannot spawn a local process — ChatGPT connectors, Claude on the web, Cursor, LibreChat — cannot start ntfy-mcp the way Claude Code does. mcp-hub is the bridge: one container serves many stdio MCP servers over Streamable HTTP, with an OAuth 2.1 login behind a single password and long-lived tokens for the clients that cannot do OAuth. Its /hub endpoint puts every server behind six meta-tools, so one connector reaches all of them without N×tool schemas in the model's context, and it speaks both protocol revisions — a question this server asks travels through it to the person at the far end instead of ending at the gateway.

Its configuration is Claude Code's mcpServers format, so the entry you already have is the entry it takes: Through mcp-hub.

Released under the MIT License.