havingautism/dsh-deepresearch
61Last commit Aug 15, 2026
dsh-deepresearch DSH plugin
Deep Research brings a structured, evidence-first research workflow to DSH. It allows users to create projects, define research questions, success criteria, and manage evidence collection. The plugin provides a dedicated web workbench for managing research projects end-to-end.
How to install the dsh-deepresearch DSH plugin
dsh plugin --profile web add github:havingautism/dsh-deepresearchCopying does not run this command. Review the repository and version before installing the dsh-deepresearch DSH plugin.
dsh-deepresearch DSH plugin data source
dsh-deepresearch DSH plugin snapshot date: Aug 16, 2026
discovered
What the dsh-deepresearch DSH plugin can do
- Record research questions, goals, constraints, seed materials, and research depth for each project.
- Edit sub-questions, dependencies, and explicit success criteria before confirming the plan.
- Refuse to write evidence until the plan is confirmed, ensuring structured investigation.
- Associate arguments, excerpts, URLs, confidence, and covered criteria with each sub-question.
- Track coverage, search and crawl budgets, limitations, and partial completion status.
- Save conclusions and final reports, with explicit marking of incomplete items.
- Search, filter, sort, resume, abort, or delete projects in the web repository.
Where the dsh-deepresearch DSH plugin fits
- Conducting a deep research project with multiple sub-questions and evidence sources.
- Managing a complex research plan that requires confirmation before collecting evidence.
- Tracking progress and coverage across different aspects of a research topic.
- Collaborating on research by using the shared web workspace and persistent state.
Who the dsh-deepresearch DSH plugin is for
- Researchers who need a structured, evidence-first workflow within DSH.
- DSH users who want to combine web and subagent capabilities for systematic investigation.
- Anyone managing multi-step research tasks that require traceability and audit trails.
dsh-deepresearch DSH plugin limitations
- The plugin does not schedule searches itself; it relies on installed Web or subagent capabilities for actual investigation.
- Search and crawl budgets are metadata only; automatic deduction requires integration with corresponding capability providers.
- Evidence is append-only and cannot be edited; incorrect claims should be re-investigated in a new project or the original project deleted.
- The plugin is still in early development; some features like automatic budget deduction are not yet implemented.
dsh-deepresearch DSH plugin: from the repository README
Quoted from the havingautism/dsh-deepresearch README, the upstream source of the dsh-deepresearch DSH plugin. Copyright remains with the original authors.
`@deepseek-ai/dsh-deepresearch` 把证据优先的 Codemini 研究工作区带到 DSH。它提供持久工作流状态、模型工具、生成的 `deepResearch` Remote namespace 和“深度研究”Web 工作区，同时组合宿主已有的 Web 与 subagent 能力。 ## ✨ 特性 - 🧭 记录研究问题、目标、约束、种子材料和研究深度。 - 🧩 确认前编辑子问题、依赖关系和明确的成功标准。 - ✅ 未确认计划时拒绝写入证据。 - 🔎 为每个子问题关联论点、摘录、URL、置信度和已覆盖标准。 - 📊 跟踪问题覆盖度、搜索与抓取预算、局限和部分完成状态。 - 📝 保存结论，以及完整或明确标记为未完成的最终报告。 - 🗂️ 在 Web 资料库中搜索、筛选、排序、恢复、中止或删除项目。 ## 🚀 快速开始 安装插件： ```sh dsh plugin --profile web add github:havingautism/dsh-deepresearch dsh web ``` 打开“深度研究”标签页，创建项目、审阅计划，并在调查前确认。插件 patch 显式设置项目、问题、标准、证据和报告上限。 ## 模型体验 ### System prompt #### What the model sees 插件激活期间，每个请求都会收到下面的研究工作流指引。 ##### 研究工作流指引 ```markdown For explicit deep research, create or resume a project before investigation. Refine and confirm its question plan, then use the composed Web and subagent tools. Save each source-backed claim against its sub-question and success criteria. Mark coverage honestly, retain limitations, and save the final report only after comparing accepted evidence. Never invent sources, evidence, coverage, or completion. ``` #### Token effect 插件激活期间每个请求承担固定的小额输入成本。 #### KV Cache 影响 文本和注册作用域不变时，本节保持前缀稳定。 ### Native 工具 #### What the model sees 模型会看到 `deep_res
Read the full READMEThe repository declares no license. Check with the authors before using it.
dsh-deepresearch DSH plugin questions
How do I install the Deep Research DSH plugin?
Run `dsh plugin --profile web add github:havingautism/dsh-deepresearch` followed by `dsh web` to start the web interface. This will add the plugin and make the deep research features available.
How do I start a new research project?
Open the 'Deep Research' tab in the DSH web interface. Create a new project by defining the research question, goals, constraints, and seed materials. Then refine the plan and confirm it before starting investigation.
Can I edit evidence after it's added?
No, evidence is append-only and cannot be edited. If you need to correct a claim, create a new project to re-investigate or delete the original project. This design ensures audit integrity.
Does the plugin perform automated web searches?
The plugin does not schedule searches itself. It relies on other installed Web or subagent capabilities (e.g., from DSH) to perform actual searches. The plugin only manages the research workflow and evidence.
How do I delete a research project?
In the web repository, you can search, filter, and sort projects. Use the delete action to remove a project entirely. Note that this will permanently remove all associated evidence and reports.