checkin_clientLABSCO SUMMARY
Built for a trusted local network, and it says so.
Client ids act as the only credential: there is no authentication beyond them, no message encryption, no access control and no audit log, which the project states plainly rather than leaving you to discover. That makes it a good fit for a local team or a set of agents on one machine, and a poor one for anything crossing an untrusted network. Within those bounds the messaging design is sensible — fire-and-forget plus an explicit collect, so nothing blocks a session waiting on a human.
What it is
A stateless HTTP Streamable MCP server that relays messages between clients. Instead of the usual client-to-server shape, it makes client-to-server-to-client work: each project declares an id and the recipients it may reach, and messages are queued per recipient until they are collected.
What you get
- Fire-and-forget sending to one or many recipients at once, with a list of messages that can differ per recipient — it returns instantly rather than blocking on a reply —
send_message_without_waiting - Collection of whatever has come back for you, which is the required second half of the pattern —
get_messages - Presence: announce that you are around and what you are working on, and see which connections are currently active —
checkin_client,get_active_sessions - A setup helper that reports your configuration back to you when the wiring is not behaving —
get_my_identity - A recipients file per project, naming your own id and the projects you can address, which your IDE's own AI agent can generate for you
- The same tools for clients that are not IDEs: LangChain, mcp-use and custom apps pass their configuration as a
recipients_configparameter instead of reading the file, with no registration step and no separate REST surface
Requirements
No account and no key — and that is also the security model, so read the next paragraph before exposing it. Run it with Python: the messaging module serves the MCP endpoint over HTTP Streamable on port 8111. The README's Docker command expects an mcp-messaging-server image that is not on Docker Hub, so that route means building the image from the repository's Dockerfile first. Python 3.11+. Each project needs a mcp_recipients.json in its root plus whatever MCP configuration its own IDE expects. The project is mcp-ide-bridge (version 1.0.0).
Setup effort
Build from source — git clone https://github.com/Mvp2o-ai/mcp-ide-bridge.git
USE CASES
01
Asking a teammate's IDE for a review
Send the request to their project id from inside your own editor, keep working, and collect the reply when it arrives.
02
Coordinating two agents on one workflow
Have the agent that finishes a phase announce it to the others, and have each check its own queue rather than polling a shared file.
03
Splitting front end and back end across two editors
Give each project its own id and recipient list, so the API side and the UI side can pass contracts and questions directly.
04
Bridging a LangChain app into the same conversation
Wrap the client so it injects its recipients configuration as a parameter, and it participates with exactly the same tools as the IDEs.
5 tools
Listed in project docs. Your client may show a different list.
send_message_without_waitingget_messagesget_my_identityget_active_sessionsBEFORE YOU INSTALL
What to know first
Tools you get
5 tools named in the project's documentation. send_message_without_waiting is the only messaging method; recipients_config is a parameter, not a tool.
Where it runs
Both documented clients run the server on your machine.
INSTALLATION
Connect your client
This one is built from source rather than installed with a single command.
Server setup
TERMINAL
# Clone and setup
git clone https://github.com/Mvp2o-ai/mcp-ide-bridge.git
cd mcp-ide-bridge
# Create and activate virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Install package in editable mode (REQUIRED for Python to find mcp_messaging module)
pip install -e .
# Run server
python -m mcp_messaging.server --port 8111MCP Clients
Cursor
CURSOR · JSON
{
"mcpServers": {
"messaging-server": {
"url": "http://localhost:8111/mcp/",
"type": "streamable-http",
"description": "MCP HTTP Streamable messaging server for client-to-client communication"
}
}
}Windsurf
Open Windsurf settings → MCP configuration; add server URL:
WINDSURF · URL
http://localhost:8111/mcp/
