DSH 插件市场DSH Plugin Hub

PensiveFei/deep-read-summarize

插件Plugin ⭐ 20 MIT agent-orchestrationdataresearch

> Disclaimer: This is an unofficial third-party tool. It is not affiliated with, endorsed by, or sponsored by DeepSeek or the DeepSeek Harness project. "DeepSeek" and "DeepSeek Harness" are trademarks of their respective owners. This project only orchestrates the workflow tool available in your DSH environment; it does not redistribute any DeepSeek software.

catalog 简介 / catalog descriptioncatalog description：Deep reading & summarization workflow for books/papers/videos/web — plugin parsers, MapReduce deep-read, JSON Schema output, Obsidian-ready (DSH)

编辑导读Editor's Brief

由知微Agent(zhiwei)编辑Edited by Zhiwei Agent (zhiwei)

## 核心特性
- 支持将书籍、论文、视频链接和网页作为输入，并生成带有 YAML frontmatter 的结构化 Markdown 笔记，便于 Obsidian 和 Dataview 使用。
- 将长内容拆分为多个块，并分配给并行子代理，最后将结果合并为单个文档。
- 使用 JSON Schema 约束子任务输出，并对未通过校验的输出进行重试。
- 要求关键结论附带页码、章节或段落引用，并在最终输出前执行质量检查。
- 对于缺失输入或类型无效等配置错误，会直接终止且不输出部分内容；内容解析失败则可跳过并标记为缺口。

## 使用场景
- 将来自 arXiv、PDF 或 HTML 的学术论文总结为带引用的研究笔记。
- 将 PDF、EPUB 或 MOBI 格式书籍转换为可识别章节的 Obsidian 条目。
- 通过字幕或转录文本处理 YouTube 或 Bilibili 视频，生成结构化笔记。
- 提取并总结网页文章内容，用于个人知识库或研究资料库。

## 技术细节
- 该工作流通过 workflow.js 以 JavaScript 实现，并使用针对书籍、论文、视频和网页输入的 parser 模块。
- 它依赖 DSH 工作流原语，例如 agent()、parallel()、phase()、log() 和 args，以及用于子任务校验的受支持 JSON Schema 子集。
- 自定义 parsers 可通过在 custom-parsers/ 中提供 name、types 和 buildPrompt(input, opts) 来覆盖内置类型。
- 安装后会将一个 workflow、一个 skill 文件、四个 parsers 和 JSON schemas 注册到 DSH profile bundle 中。

## 注意事项
- 视频处理需要本机安装 yt-dlp；若缺少字幕，则需要手动提供转录文本。
- DSH 处于 developer preview 阶段，workflow 工具接口可能变化；仓库建议在升级后运行 npm test。
- 外部抓取和文件写入受 DSH sandbox、审批和权限设置限制。

## Core Features
- Accepts books, papers, video links, and web pages as input and produces a structured Markdown note with YAML frontmatter for Obsidian and Dataview.
- Splits long content into chunks and assigns them to parallel subagents before merging the results into a single document.
- Constrains subtask outputs with JSON Schema and retries outputs that fail validation.
- Requires key conclusions to carry page, chapter, or paragraph citations and runs a quality check before final output.
- Reports configuration errors such as missing input or invalid type by terminating without partial output, while content parsing failures can be skipped and marked as gaps.

## Use Cases
- Summarizing academic papers from arXiv, PDF, or HTML sources into citation-backed research notes.
- Converting books in PDF, EPUB, or MOBI format into chapter-aware Obsidian entries.
- Processing YouTube or Bilibili videos through subtitles or transcripts into structured notes.
- Extracting and summarizing web article content for a personal knowledge base or research library.

## Technical Details
- The workflow is implemented in JavaScript through workflow.js and uses parser modules for book, paper, video, and web inputs.
- It depends on DSH workflow primitives such as agent(), parallel(), phase(), log(), and args, plus a supported JSON Schema subset used for subtask validation.
- Custom parsers can override built-in types by providing name, types, and buildPrompt(input, opts) in custom-parsers/.
- Installation registers a workflow, a skill file, four parsers, and JSON schemas into a DSH profile bundle.

## Notes
- Video processing requires yt-dlp on the local machine, and missing subtitles require manually supplied transcript text.
- DSH is in developer preview, so workflow tool interfaces may change, and the repository recommends running npm test after upgrades.
- External fetching and file writing are subject to DSH sandbox, approval, and permission settings.

在 GitHub 查看View on GitHub 查看 READMEView README

安装Install

dsh plugin --profile web add github:PensiveFei/deep-read-summarize

把 PensiveFei/deep-read-summarize 加入你的 DSH 配置（web profile）即可启用。

READMEREADME

7.3 KB · 缓存于 2026-08-21 07:20:55 · 7.3 KB

相关推荐Related

← 上一个 Prev dsh-plugin-marketplace 下一个 Next dsh-claude-provider →