{"url":"https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2","title":"Claude Skill Extracts Messy Docs to Reports","domain":"medium.com","imageUrl":"https://images.pexels.com/photos/37021497/pexels-photo-37021497.jpeg?auto=compress&cs=tinysrgb&h=650&w=940","pexelsSearchTerm":"claude","category":"Tech","language":"en","slug":"b85bc7f0","id":"b85bc7f0-7ce5-42e4-9a17-83fab8471e19","description":"Claude Skill Built: Umair Ali Khan created a custom Claude Skill paired with an MCP server to process messy documents and media into structured reports.[[1","summary":"## TL;DR\n- **Claude Skill Built:** Umair Ali Khan created a custom Claude Skill paired with an MCP server to process messy documents and media into structured reports.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)[[2]](https://medium.com/@umairali.khan/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n- **Supported Formats:** Handles .docx, .pdf, .ppt, .xlsx files plus images and audio/video recordings for knowledge extraction.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n- **Local Setup Guide:** Article teaches quick local installation and workflow integration using the MCP tool.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\n## The story at a glance\nUmair Ali Khan, a senior AI researcher, describes building a Claude Skill that uses an MCP server to extract information from diverse document formats and generate reports matching a user template. The piece is a how-to tutorial for knowledge workers facing scattered data across files like documents, spreadsheets, presentations, images, and media. It appears amid growing use of Anthropic's Claude features like Skills and Model Context Protocol (MCP) for specialized AI tasks.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)[[2]](https://medium.com/@umairali.khan/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\n## Key points\n- Custom Claude Skill combines with MCP server to pull key info from multiple file types including .docx, .pdf, .ppt, .xlsx, images, audio, and video.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n- Generates structured reports based on a provided template, addressing common challenges for organizations and knowledge workers.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n- Includes practical example of instant extraction and report creation from piles of messy, multi-format documents.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n- Provides step-by-step instructions for local setup of the Skill and MCP server, such as checking templates and sample documents.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n- Aimed at routine workflow integration to save time on report writing from scattered sources.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\n## Details and context\nThe article targets data professionals dealing with information spread across formats, a frequent task in organizations. Claude Skills are reusable instructions that guide Anthropic's Claude AI for specific workflows, while MCP (Model Context Protocol) acts as a tool for external interactions like file processing.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\nSetup involves running an MCP server locally as a tool within the Claude Skill, enabling handling of multimodal inputs without manual sorting. Khan's background in AI/ML, including LLMs, RAG, MCP, and knowledge extraction, informs the tutorial.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\nNo full code or exact steps are visible due to paywall, but it emphasizes quick local implementation over cloud dependencies.[[2]](https://medium.com/@umairali.khan/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\n## Key quotes\n- \"This article presents a practical example of instantly extracting the required information from a pile of documents scattered across multiple formats (.docx, .pdf, .ppt, .xlsx, images, audio/video recordings) and generating a structured report using a given template with a custom Claude Skill using an MCP server as a tool.\" – Article TL;DR[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\n## Why it matters\nClaude Skills and MCP expand AI from general chat to specialized tools for real work like report generation, reducing manual data handling in knowledge-intensive fields. Readers in data science or consulting can adopt this for faster analysis of mixed media, potentially cutting hours from routine tasks. Watch for Anthropic updates to Skills or MCP, as they could add more file types or cloud options, though local setup remains key for privacy.\n\n## FAQ\nQ: What file formats does the Claude Skill process?\nA: It extracts from .docx, .pdf, .ppt, .xlsx, images, and audio/video recordings. The tool handles piles of these to pull required information for reports. This multimodal support fits scattered organizational data.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\nQ: How does the MCP server fit into the Claude Skill?\nA: The MCP server serves as a tool within the custom Claude Skill for processing documents and media. It enables extraction and report generation based on templates. Local setup integrates it into Claude workflows.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\nQ: What does the article teach about setup?\nA: It shows quick local setup of the Skill and MCP server for routine use. Steps include checking templates and sample documents. The goal is easy workflow integration without complex coding.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)\n\nQ: Who is the article for?\nA: Knowledge workers and organizations writing structured reports from multi-format documents. It addresses common pain points in analyzing scattered info. The tutorial suits AI/ML practitioners familiar with Claude.[[1]](https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2)","hashtags":["#ai","#claude","#skills","#mcp","#knowledge","#extraction"],"sources":[{"url":"https://medium.com/data-science-collective/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2","title":"Original article"},{"url":"https://medium.com/@umairali.khan/i-created-a-claude-skill-that-turns-piles-of-messy-documents-media-into-a-structured-report-19e9950f93b2","title":""}],"viewCount":2,"publishedAt":"2026-04-23T23:11:14.393Z","createdAt":"2026-04-23T23:11:14.393Z","articlePublishedAt":"2026-02-01T18:58:03.472Z"}