Exstruct by Harumiweb is an mcp tool designed for AI agents and LLM/RAG pipelines that process Excel workbooks. Written in Python, it serves as an AI Agents skill and an open-source CLI for managing Excel data. The program includes an MCP server, acting as an integration layer for host-managed systems. It enables AI coding assistants to interact with Excel, converting complex spreadsheet structures into usable JSON. Exstruct supports a controlled editing workflow, ensuring data integrity for automated processes. Its capabilities extend to both reading and writing Excel content for various AI-driven applications.

What does Exstruct connect to?

Exstruct connects directly to Excel workbooks, serving as a critical component for AI Agents skills and LLM/RAG pipeline development. It functions as an MCP Open Source tool, providing an MCP server. This server acts as an integration and compatibility layer around its core editing functions, allowing for host-managed integrations. The tool is compatible with AI clients such as Claude Code, Codex CLI, and ChatGPT, using the stdio transport protocol. This setup allows AI systems to interact with Excel data effectively.

Core features and tools

Exstruct provides several callable tools for Excel interaction, including exstruct_extract for data retrieval and exstruct_capture_sheet_images. It also offers exstruct_make and exstruct_patch for modifying workbooks, alongside exstruct_read_json_chunk. The command-line interface exposes commands like "patch", "make", "validate", "ops list", and "ops describe". Additionally, an "exstruct-cli" skill is available for agents. These tools collectively support a range of operations for managing and transforming Excel data within AI workflows.

Who is it for?

Exstruct targets developers building LLM and RAG pipelines that require ingesting Excel data. It is also suitable for AI coding assistants and teams seeking transparency over black-box tools. Organizations comfortable with maintaining internal forks will find value in its open-source nature. The tool supports scenarios where autonomous Excel reading and writing by AI agents are needed. Skip this if your project does not involve AI agents or LLM/RAG pipelines interacting with Excel workbooks.

How to get started

No credentials are required for Exstruct's direct operation, as its MCP server is designed for host-managed integrations. The server supports the stdio transport protocol.

Use cases

• Conversion from Excel to structured JSON, including tables, shapes, and charts, for LLM/RAG pipelines.
• Autonomous Excel reading and writing by AI agents through its CLI and MCP integration.
• A safe editing workflow for Excel workbooks, involving steps like "validate", "dry-run", "inspect", "apply", and "verify".

Exstruct offers a specialized solution for integrating Excel data into AI-driven applications and LLM/RAG pipelines. Its open-source nature and MCP server capabilities allow for deep integration with various AI clients, including Claude Code and ChatGPT. The tool helps manage Excel content through a set of callable functions and CLI commands, supporting structured data extraction and controlled editing. Try converting an Excel workbook to structured JSON using the "exstruct_extract" tool for your LLM/RAG pipeline.