nicolascroce/keepsake-mcp

Overview

An MCP server that bridges an AI agent with Keepsake, a personal CRM. It exposes contacts, tasks, and notes as resources and tools that the agent can read and write through the Model Context Protocol. Built in TypeScript, it runs as a standalone process that connects to Keepsake's data store.

Best for

Best for
Developers building AI workflows around personal relationship management

Use cases

• Automatically log meeting notes and follow-ups to a contact
• Query tasks for a specific person or project
• Update contact details or add notes via a chat interface

How to use

Install

claude mcp add keepsake -- npx -y keepsake-mcp

Tools exposed

• list_contacts
• get_contact
• create_contact
• update_contact
• delete_contact
• search_contacts
• get_contact_timeline
• list_companies
• get_company
• create_company
• update_company
• delete_company
• search_companies
• list_entries
• create_entry
• update_entry
• delete_entry
• list_tasks
• create_task
• update_task

Tested with

Claude Desktop, Claude Code, Cursor

Example client config

{\n  "mcpServers": {\n    "keepsake": {\n      "type": "streamable-http",\n      "url": "https://app.keepsake.place/api/mcp",\n      "headers": {\n        "Authorization": "Bearer ksk_YOUR_API_KEY"\n      }\n    }\n  }\n}

Notes

An MCP server that bridges an AI agent with Keepsake, a personal CRM. It exposes contacts, tasks, and notes as resources and tools that the agent can read and write through the Model Context Protocol. Built in TypeScript, it runs as a standalone process that connects to Keepsake’s data store.

0 stars on GitHub. Last updated 2026-04-16. Licensed MIT.

Use cases

• Automatically log meeting notes and follow-ups to a contact
• Query tasks for a specific person or project
• Update contact details or add notes via a chat interface

Pros

• Direct integration between AI agents and personal CRM data
• Leverages the standardized MCP protocol for interoperability
• Open source and easy to inspect or extend

Cons

• Zero stars on GitHub suggests very early stage or limited adoption
• Requires a running Keepsake instance and MCP-compatible client
• Dependency on third-party Keepsake service may affect reliability

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.