get_chunkLABSCO SUMMARY
Big files answered in pieces instead of refused.
The working order is the point: preview first to learn the columns and the row count, then either page through in chunks or, for JSON, search the file directly and get back only the matching entries with a count of what was scanned. Because the JSON tools stream, the file's size stops being the constraint and the response size becomes the thing you control. Pick the transport by where the files live — the hosted endpoint cannot see your disk.
What it is
A reader for .xlsx, .csv and .json files built around not returning everything at once. Previews come back with row counts and column names; the rest is fetched in chunks or searched in place, with JSON streamed rather than parsed into memory.
What you get
- An Excel or CSV file opened as a preview of the first 100 rows plus the total row count and column list, or in full when the file is small enough
- Rows fetched in chunks from a start index, 1000 at a time by default, with only the columns you name
- The same preview shape for JSON — the first 100 entries plus the total count and field list, with anything over 1000 entries treated as large
- A streaming search over a JSON file by field, with contains, equals, startsWith or endsWith, returning matches, the match count and how many entries were scanned, up to 1000 results
- JSON entries fetched in chunks by start and limit for iterative passes over an entire dataset
- Field and column filtering on every read, so a wide file can be narrowed to the two columns the question is about
Requirements
Node.js 18 or newer and nothing to authorize. Over stdio it runs as npx -y excel-analyser-mcp and reads absolute paths on your own machine. The same package serves Streamable HTTP with npx excel-analyser-mcp streamableHttp, default http://localhost:8080/mcp, and the deprecated SSE transport with the sse subcommand; a port and endpoint can follow either. A ready-made hosted HTTP endpoint exists for zero-install use, but it runs elsewhere, so it can only reach files at public URLs — local paths only work on a server running on your machine. ISC.
Setup effort
One command — npx -y excel-analyser-mcp
USE CASES
01
Working out what is in a spreadsheet you were just handed
Open it for the column names, the total row count and a first look at the data, without pulling the whole file into the conversation.
02
Processing a large export in passes
Fetch rows a thousand at a time from a start index, narrowed to the columns that matter, and walk the file to the end.
03
Finding a few records inside a huge JSON file
Search by field with contains or equals; the file is streamed and only the matches come back, with a count of how many entries were scanned to find them.
04
Reading every entry of a dataset that will not fit
Preview it for the field list and total, then page through the entries by start and limit until the count runs out.
5 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
get_json_chunkquery_jsonread_excelread_jsonBEFORE YOU INSTALL
What to know first
Tools you get
5 tools when we connected.
Requires
Node.js (v18 or higher recommended)
Where it runs
Both documented clients run the server on your machine.
CONTEXT COST
Light Standing cost
Rated by the size of the 5 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
npx
TERMINAL
npx -y excel-analyser-mcpHosted endpoint
ENDPOINT URL
https://web-production-64851.up.railway.app/mcpAdd this hosted endpoint to your MCP client.
MCP Clients
Claude Desktop, Cursor
JSON
{
"mcpServers": {
"Excel Analyser MCP": {
"command": "npx",
"args": [
"-y",
"excel-analyser-mcp"
]
}
}
}
