conversation-handoff
MCP server for transferring conversation context between AI chats and projects
claude mcp add --transport stdio trust-delta-conversation-handoff-mcp npx conversation-handoff-mcp@latest
How to use
The conversation-handoff MCP server enables cross-chat context transfer by providing a lightweight in-memory store for handoff data that can be accessed by MCP clients. It supports saving full conversation exchanges, listing and loading handoffs, and merging multiple handoffs into a single context. Tools exposed by this MCP server include handoff_save for persisting a conversation, handoff_list to see summaries of saved handoffs, handoff_load to retrieve full content, handoff_clear to delete stored handoffs, and handoff_merge to combine several handoffs into one consolidated context. This server is designed to work with MCP clients such as Claude Desktop, Claude Code, Codex CLI, Gemini CLI, and others that implement the MCP protocol, enabling seamless transfer of user-driven context between chats or projects.
How to install
Prerequisites:
- Node.js and npm installed on your machine.
- Basic familiarity with running MCP servers via npx or npm.
Option A: Run via npx (recommended, no pre-installation required)
- Ensure you have an internet connection.
- Start the MCP server using npx: npm run --silent -- npx conversation-handoff-mcp@latest
Option B: Global installation (optional if you want a persistent executable)
- Install the package globally: npm install -g conversation-handoff-mcp
- Run the server via the generated CLI (if exposed) or via npx equivalent depending on your setup.
Option C: Local development build (for contributors)
- Clone the repository: git clone https://github.com/trust-delta/conversation-handoff-mcp.git
- Install dependencies: cd conversation-handoff-mcp npm install
- Build (if a build step is provided): npm run build
- Run locally (example, adjust path to built index if necessary): node dist/index.js (or the appropriate entry point defined by the project)
Note: The MCP configuration can be used by clients as shown in the example in the repository to connect to the server.
Additional notes
Tips and known considerations:
- The server stores data in memory; data will be lost if the server process stops. Use it for active session sharing rather than long-term persistence.
- If you enable audit logging or other features, ensure appropriate log routing and opacity for sensitive data.
- When using Auto-Connect modes in MCP clients, ensure your HANDOFF_SERVER setting does not override your intended target during testing.
- For Codex CLI and other TOML-based clients, remember to adapt the configuration format if required by the client (the MCP server itself uses the standard JSON config described here).
- If you run into port conflicts, configure the server to listen on a specific port if the MCP implementation allows; otherwise let the client auto-discover as designed.
Related MCP Servers
Remote
A type-safe solution to remote MCP communication, enabling effortless integration for centralized management of Model Context.
mcp-typescript-sdk
A TypeScript SDK for implementing Model Context Protocol (MCP) over MQTT, supporting both browser and Node.js environments.
pluggedin-app
The Crossroads for AI Data Exchanges. A unified, self-hostable web interface for discovering, configuring, and managing Model Context Protocol (MCP) servers—bringing together AI tools, workspaces, prompts, and logs from multiple MCP sources (Claude, Cursor, etc.) under one roof.
marinade-finance
Marinade Finance MCP Server is an MCP server specifically designed for the Marinade Finance.
google-scholar
An MCP server for Google Scholar written in TypeScript with Streamable HTTP
mcp -blog
Example of a MCP implementation using TypeScript and OAuth.