initialize_arrayLABSCO SUMMARY
The list gets finished, not abandoned halfway.
Asked to process fifty items, a model will do eight and summarise. Handing them out one at a time with the remaining count attached — and refusing to release the results until the array is empty — turns that into a loop that actually terminates.
What it is
A processing harness for a list: load an array with a task description, pull items one at a time or in batches, store each result as it comes, and take the whole set once every item is done.
What you get
- An array loaded with a task description and an optional batch size, defaulting to 1
- The next item returned with its index, the task, and how many remain
- The next batch, when a batch size is set, with indices and remaining count
- A result stored per item or per batch
- All results retrieved at the end with an optional summary — the call errors while items remain
- A reset that clears the current processing state
Requirements
Node, with npm install and npm start. No account and no key, and nothing leaves the machine.
Setup effort
Build from source — git clone https://github.com/smogili1/loop_mcp.git
USE CASES
01
Applying the same treatment to a long list
Load the array with the instruction once, and let each item come back with the task attached.
02
Making sure nothing was skipped
Retrieve the results only at the end — the call refuses while items are outstanding.
03
Trading calls for throughput
Set a batch size so items arrive several at a time instead of one per round trip.
04
Keeping results attached to their input
Store each result as it is produced rather than holding fifty of them in the conversation.
05
Wrapping up with an overview
Ask for the summary alongside the full result set.
6 tools
Listed in project docs. Your client may show a different list.
get_next_itemget_next_batchstore_resultget_all_resultsresetBEFORE YOU INSTALL
What to know first
Tools you get
6 tools named in the project's documentation.
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
You clone this one yourself.
Server setup
TERMINAL
git clone https://github.com/smogili1/loop_mcp.git
cd loop_mcp
npm installMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"loop-processor": {
"command": "node",
"args": [
"<YOUR_PATH_TO_LOOP_MCP>/server.js"
]
}
}
}
