home / mcp / threads mcp server
Exposes Threads CLI data and operations via the Model Context Protocol for programmatic access.
Configuration
View docs{
"mcpServers": {
"joshuaramirez-threads-mcp": {
"command": "node",
"args": [
"dist/server.js"
]
}
}
}The Threads MCP Server exposes Threads CLI data and operations through the Model Context Protocol, letting you manage threads, containers, groups, and progress from MCP clients. It supports standard MCP tooling and integrates directly with the local Threads data store, enabling scripted workflows, automation, and CLI-style interactions from your preferred MCP client.
To use the Threads MCP Server, first install and build the server, then run it as a local process that MCP clients can connect to. You interact with the server by issuing MCP actions through your client to manage threads, containers, groups, and progress. You can also inspect the server during development to verify responses and behavior.
Prerequisites: Node.js and npm must be installed on your system.
Clone or download the project, then install dependencies and build the server.
cd threads-mcp
npm install
npm run buildRun the server as a local process that MCP clients can connect to. The server is started by Node and uses the compiled distribution.
node dist/server.jsConfigure your MCP client to connect to the local server as a stdio MCP server. Use the command and path that correspond to your environment. The server can be started with the runtime command shown above, and your client will communicate over the standard MCP transport.
Data is stored in the Threads data location and is compatible with the Threads CLI. The MCP server reads and writes to the same data store, so operations performed through MCP mirror CLI behavior.
Create and track a project: use create_container to create a project container, create_thread to add threads under that container, log updates with add_progress, and use get_next_action to identify the next item to work on.
Review active work: view threads://threads/active or list_threads with a status filter, then inspect progress with list_progress.
Organize existing threads: create groups, move threads into groups, and establish hierarchy with set_parent.
Keep your environment secure by restricting access to the local MCP server and protecting any tooling that interacts with your Threads data store. When deploying in shared environments, consider running the MCP server behind authentication and using scoped permissions for clients.
If you are developing or testing, you can also use the MCP Inspector to validate responses by running the server with a wrapper that exercises MCP endpoints.
Create a new thread within a container or project.
Update properties of an existing thread.
Archive a thread, optionally cascading the action to related entities.
Permanently delete a thread.
Retrieve a thread by ID or name.
Add a progress note to a thread.
Retrieve the progress history for a thread.
Edit an existing progress entry.
Delete a progress entry.
Create an organizational container.
Update container properties.
Delete a container.
Create an organizational group.
Update group properties.
Delete a group.
Set the parent of an entity.
Move an entity into a group.
Get immediate children of an entity.
Get the path from an entity to the root.
Get the full subtree under an entity.
List threads with optional filters.
List containers with filters.
List all groups.
Perform a full-text search on threads.
Suggest the next action for a thread.
Get the complete hierarchy tree.
Get any entity by ID or name.
Get a container by ID or name.
Get a group by ID or name.