excel-parser
An open-source Python library that parses Excel workbooks into citation-ready, token-counted JSON chunks for language models, retrieval-augmented-generation pipelines and AI agent frameworks, preserving cells, formulas, merged regions, tables and charts.
excel-parser converts an .xlsx file into a structured, typed JSON representation that keeps information a large language model would otherwise lose by reading a spreadsheet directly: cell values, formulas, merged regions, tables, charts, conditional formatting and dependency graphs. Each output chunk is tied back to an exact sheet-and-range reference so an answer generated from it can cite the source cells, and chunks are sized by token count for direct use in LangChain, LangGraph, CrewAI, the OpenAI Agents SDK, or any MCP-aware client such as Claude Desktop, Cursor, Windsurf or Zed.
On a public benchmark of hundreds of real-world spreadsheet tasks, the project reports parsing thousands of files with a small number of timeouts and no errors, and a close but favorable comparison against another spreadsheet parser on retrieval recall at several cutoffs, with faster median parse times per file.
It is written in Python and released under the MIT license.