MCP server · Memtro

Memtro is a Model Context Protocol server. Any MCP client that supports Streamable HTTP can connect and use your memory tools and every connector tool you have access to, as you.

URL: https://platform.memtro.com/api/mcp
Auth: Authorization: Bearer sk-memtro-… (an API key from Dashboard → API keys)
Transport: Streamable HTTP, stateless

Claude Code

claude mcp add --transport http memtro https://platform.memtro.com/api/mcp \
  --header "Authorization: Bearer sk-memtro-..."

Then, inside Claude Code: "search memtro for who owns billing", "use memtro to look up Help Scout ticket 2450". Full guide: Claude Code.

OpenCode

opencode.json:

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": { "Authorization": "Bearer sk-memtro-..." }
    }
  }
}

Full guide: OpenCode.

Cursor, Claude Desktop and others

Add an MCP server of type HTTP with the URL and header above. Tool names are memtro_search, memtro_remember, memtro_explore, the job tools, and <connector>_<action> for your connections (for example hubspot_search, postgres_query, helpscout_get_conversation).

What happens on the server

Tool calls over MCP behave exactly like tool calls inside a Memtro chat: results are cached briefly, connector results are anchored into the memory graph, and facts you save with memtro_remember are stored under the scope you choose. The MCP endpoint is also what the OpenCode and Claude Code engines use internally, with per-request tokens instead of API keys.