AI Switch MCP Server
Enables session continuity across AI coding assistants by preserving context and generating handoff markdown, allowing seamless switching between tools like Cursor, Claude Code, and Claude Desktop.
README
AI Switch MCP Server
Open-source MCP server for session continuity across AI coding assistants.
When one assistant runs out of credits, another one can continue with full context from the same persistent session store.
What This Solves
- Preserve coding context between Cursor, Claude Code, Claude Desktop, and other MCP clients.
- Generate clean handoff markdown + resume prompts.
- Keep a timeline of progress, files touched, current step, and next steps.
- Store sessions in one shared location so every client sees the same state.
Included MCP Tools
continue_task- one-call auto-continue tool for prompts like "continue auth feature".start_session- create a new handoff session.append_progress- append assistant progress to a session.update_session- patch task metadata, notes, status, files, and optional progress.get_session- fetch a session by ID (with optional handoff text).get_latest_session- get most recently updated session for a project.list_sessions- list recent sessions with filters.complete_session- close out a session.generate_handoff- produce portable handoff markdown + resume prompt.delete_session- delete a session (requiresconfirm=true).store_info- show the active session store path.
Storage
By default, sessions are stored in:
~/.aiswitch/sessions.json
You can override this (recommended when sharing across multiple clients/sandboxes):
AISWITCH_STORE_DIRAISWITCH_STORE_FILE
Local Development
npm install
npm run build
npm test
Run server over stdio:
node dist/src/index.js
Cursor Setup
Use global ~/.cursor/mcp.json or project .cursor/mcp.json:
{
"mcpServers": {
"ai-switch": {
"command": "node",
"args": ["/ABSOLUTE/PATH/TO/AISwitchMCPServer/dist/src/index.js"],
"env": {
"AISWITCH_STORE_DIR": "/ABSOLUTE/PATH/TO/shared-aiswitch-store"
}
}
}
}
After saving, reload Cursor and confirm the server appears in MCP settings.
Claude Code Setup
Recommended (CLI-managed):
claude mcp add --transport stdio --scope user ai-switch -- \
node /ABSOLUTE/PATH/TO/AISwitchMCPServer/dist/src/index.js
With shared storage path:
claude mcp add --transport stdio --scope user \
--env AISWITCH_STORE_DIR=/ABSOLUTE/PATH/TO/shared-aiswitch-store \
ai-switch -- node /ABSOLUTE/PATH/TO/AISwitchMCPServer/dist/src/index.js
Project-scoped alternative: create .mcp.json in repo root.
{
"mcpServers": {
"ai-switch": {
"type": "stdio",
"command": "node",
"args": ["/ABSOLUTE/PATH/TO/AISwitchMCPServer/dist/src/index.js"],
"env": {
"AISWITCH_STORE_DIR": "/ABSOLUTE/PATH/TO/shared-aiswitch-store"
}
}
}
}
Claude Desktop Setup (macOS)
Edit:
~/Library/Application Support/Claude/claude_desktop_config.json
Add:
{
"mcpServers": {
"ai-switch": {
"command": "node",
"args": ["/ABSOLUTE/PATH/TO/AISwitchMCPServer/dist/src/index.js"],
"env": {
"AISWITCH_STORE_DIR": "/ABSOLUTE/PATH/TO/shared-aiswitch-store"
}
}
}
}
Then fully restart Claude Desktop.
Quick Continue (Minimal Prompting)
The server includes continue_task, designed for short prompts like:
- "continue the auth feature"
- "resume payment webhook work"
How it works:
- Tries explicit
session_idif provided. - Otherwise matches by
project_path(or current working directory). - Otherwise searches by your request text across recent sessions.
- Returns:
resume_prompthandoff_markdownquick_continue_instruction
In practice, tell the assistant:
Use ai-switch continue_task for: "continue auth feature", then continue implementation.
Recommended Workflow
- In current assistant, call
start_sessionat task start. - During work, call
append_progressorupdate_session. - When switching assistants, call
generate_handoff. - Paste
resume_promptinto the next assistant (or callget_latest_sessionthere). - Continue updating the same
session_id.
Open Source + GitHub Release
- Create a GitHub repository and push this project.
- Keep
README.md+LICENSE(MIT) in root. - Add CI (optional): run
npm teston PRs. - Tag releases (
v1.0.0,v1.1.0, etc.). - (Optional) Publish to npm so others can install easily.
Optional npm publish flow
# Update name if needed (must be unique on npm)
npm version patch
npm publish --access public
Then users can run it in MCP configs with:
{
"mcpServers": {
"ai-switch": {
"command": "npx",
"args": ["-y", "ai-switch-mcp-server"]
}
}
}
Security Notes
- Do not store secrets in session notes.
- Keep
AISWITCH_STORE_DIRin a location with proper filesystem permissions. - If sharing
.mcp.jsonin a repo, avoid hardcoded secret env vars.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.