高效文档处理多智能体团队:用五个 Agent 串起文档读写与报告
SkillHub 社区 Skill「高效文档处理多智能体团队」(@user_d76c672d/doc-processing-team,v1.0.3)基于 agent-team-template v2.1.0,由 5 个专业 Agent 协作完成 PDF/Word/Excel/图片的读取、英译中、分类、总结与报告生成。覆盖 PDF 转 Word/Markdown、批量归档、合同审阅、OCR、文献整理、文档问答与信息抽取等场景;声明国内可离线、不上传敏感数据,且 requires_api_key=false。本文依据 SkillHub 官方简介与安装文档说明 skillhub 安装方式及 Cursor、Claude Code、Codex 等 skills 目录要点。
Read MoreIntroduction to the DeepSeek Harness Plugin dsh-r7-office
dsh-r7-office is a DeepSeek Harness plugin or standalone MCP server designed to read, write, and edit DOCX, XLSX, PPTX, and PDF files by manipulating raw OOXML data and using a local R7-Office engine. It offers core features such as format preservation, built-in validation, and dual-mode operation, supporting use either as an integrated plugin or an independent service. R7-Office must be installed separately, and the desktop bridge feature is available only on Windows.
Read MoreDSH Plugin Introduction: dsh-doc-quick
`dsh-doc-quick` is a DSH plugin designed to address the challenge of agent access to local file systems. Users drag and drop documents in the Web interface, the host application saves them to a local directory, and the Agent reads, processes, and outputs the results in the right sidebar. The plugin supports multiple formats, including PDF, Word, and Markdown, and provides tools for file listing, reading, and completion registration. Installation is straightforward, and it relies on `pdfjs-dist` for PDF parsing. It is ideal for quickly processing local documents in a Web environment, significantly reducing interaction overhead.
Read MoreDeepSeek Harness Document Read/Write Toolkit (dsh-doc-toolkit)
dsh-doc-toolkit is a document read/write plugin for the DeepSeek Harness (DSH) framework, designed to meet the needs of AI assistants handling office documents. It supports reading and writing PDF, DOCX, XLSX, and CSV, with features including paginated processing of large files, Chinese PDF export, and AI-friendly text interfaces. The plugin can be enabled with an installation command and provides a natural language interaction interface for reading document content and generating report documents. Note that PDFs must contain a text layer to be read, and the plugin has specific compatibility requirements for DSH versions.
Read Moredsh-md-convert: A MarkItDown-Based Document-to-Markdown Converter Plugin
`dsh-md-convert` is a localized document-to-Markdown plugin in the DeepSeek Harness (DSH) ecosystem. Powered by Microsoft's MarkItDown engine, it converts Office documents and PDFs (including scanned copies) into structure-preserving Markdown. The plugin supports two invocation methods—CLI commands and Agent tools—with automatic routing: modern formats are converted directly, legacy formats require WPS/Office/LibreOffice, and scanned documents are recognized using a lightweight OCR model (with layout analysis and table extraction). Once installed, Agents can invoke `md_convert` directly, while the CLI also supports batch processing and dependency checks. Before running, the corresponding backend must be installed.
Read Morexby-atlas-docs: DSH Document Processing Service
xby-atlas-docs is a DeepSeek Harness (DSH) plugin designed to help large models more effectively leverage technical documentation. It uses an MCP server to convert official documentation into Markdown format for use by clients such as Cursor and Cline. Core features include setting API keys, listing documentation libraries, searching pages, retrieving documentation indexes, and accessing specific page content. The typical workflow is: set up an API key -> list documentation -> explore or search -> retrieve details. This plugin substantially lowers the barrier for developers referencing documentation for specific libraries and improves information retrieval efficiency.
Read Moredsh-sidebar-superdoc-docx: Edit DOCX Files Directly in the DSH Web Sidebar
The `dsh-sidebar-superdoc-docx` plugin for DeepSeek Harness (DSH) provides browser-native DOCX editing capabilities for web interfaces. It allows users to open, edit, and save Word files in the sidebar, with support for comments, tracked changes, offline operation, and synchronization with the AI context. This plugin depends on `dsh-better-sidebar`; once installed, it replaces the built-in viewer in the sidebar, enabling seamless file preview and human-AI collaboration.
Read MoreIntroduction to the DeepSeek Harness Documentation Tool Plugin
`dsh-tool-doc` is a plugin for DeepSeek Harness (DSH) designed to let Agents process documents directly without invoking external commands via the Shell. It provides three core tools—read, create, and edit—and supports formats such as PDF, DOCX, XLSX, and CSV. Editing operations support text replacement, cell updates, and appending rows, with strict validation mechanisms. Key limitations include DOCX supporting only paragraph-level replacement, PDF/PPTX not supporting in-place editing, and binary format writes not adhering to the sandbox policy. By integrating document operations, the plugin significantly reduces the latency and complexity of document processing for Agents.
Read Moredsh-office-cli: Single-Binary Office Suite
`dsh-office-cli` is a plugin for DeepSeek Harness designed to address the pain point of complex environment setup when agents control Office documents (Word, Excel, and PPT). Packaged as a single binary, it integrates path selection, a closed-loop HTML rendering process (render→look→fix), and real-time preview features, enabling lightweight AI document processing. Developers need to install it from the GitHub source and enable it in DSH. When using it, attention should be paid to permissions and security practices. It is suitable for automated document generation scenarios.
Read Moredsh-codex-media: Local OpenAI Codex-Powered Image and Document Analysis Tool
`dsh-codex-media` is a local media analysis tool designed for DeepSeek Harness (DSH), addressing the limitation of pure-text models in directly processing images and documents. The plugin provides image description, document analysis, and image generation capabilities, runs using the local OpenAI Codex CLI, and requires no third-party runtime libraries. It must be paired with the `dsh-drop-to-path` plugin before installation, supports multiple transport methods (CLI, Hermes, and API), and enforces strict permission and security validation mechanisms.
Read MoreDSH Tool Markdown: DSH Markdown Tool Plugin
`omdsh-dev/dsh-tool-markdown` is an MIT-licensed document processing plugin for DSH, designed to address format instability and security risks in HTML, Markdown, and table conversions. The plugin registers a `markdown` tool and supports four core functionalities through different `action` values: converting HTML to GFM Markdown, converting Markdown to whitelist-safe HTML, normalizing HTML or pipe-delimited text into GFM tables, and generating nested table of contents based on headings. The plugin adopts a zero-dependency, pure-function architecture with strict security boundaries: it prohibits script execution, blocks remote resource loading, and automatically strips dangerous tags such as `script` and `style`; output only allows whitelisted tags, forces text escaping, and retains only links.
Read MoreMinerU Document Parsing Plugin for DeepSeek Harness
`dsh-mineru` is an MIT-licensed plugin for DeepSeek Harness (DSH), designed to parse documents such as PDFs, Word, PPT, Excel, HTML, and images into structured Markdown for efficient processing by intelligent agents. This plugin integrates the MinerU engine and offers dual API modes: the Precision API with a configured token supports large files, batch parsing, and a wider range of Office formats; the Agent API without a configured token requires no key but is subject to IP rate limiting and supports only single and smaller files. The plugin automatically selects the appropriate mode based on configuration. Core features include progressive tool exposure, drag-and-drop upload via Web UI, automatic chunking and batch parsing, rate limit handling, and result persistence to disk. Installation requires DSH.
Read More