dsh-web
zhu1090093659/dsh-webDeepSeek Harness(DSH)Web 插件聚合生态包 · 一切皆插件,创意工坊分发(dsh-market.com)
A local literature review, PDF annotation, evidence synthesis, and research notes workbench for DeepSeek Harness.
A local-first literature workspace for psychology and social-science research
Discover papers, read and annotate PDFs, organize evidence, conduct systematic reviews, and collaborate with a DeepSeek Harness agent.
中文 · English · User guide (Chinese) · Issues
[!IMPORTANT]
0.5.0-beta.1targets the DeepSeek Harness developer preview. The full workflow has been validated on Windows; automated CI covers Ubuntu/Windows with Node.js 22 and 24.
Hana Research keeps literature discovery, PDFs, notes, screening decisions, and evidence exports in one traceable workspace:
| Stage | Capabilities |
|---|---|
| Discovery | Four-source search, AI interpretation, saved searches, new-result alerts, journal monitoring |
| Organization | Project libraries, paper roles, reading status, priorities, tasks, cross-paper relations |
| Reading | Outline/thumbnails/search, colored annotations, page-linked excerpts, reading progress, structured summaries |
| Systematic review | Inclusion/exclusion criteria, title/abstract and full-text screening, dual review, conflict resolution, PRISMA 2020 |
| Evidence synthesis | Custom extraction fields, RoB 2 / ROBINS-I, GRADE, evidence matrices, argument links |
| Export and agent work | Markdown, DOCX, PDF, CSV, XLSX, BibTeX, RIS, plus 20 Harness agent tools |
The systematic-review modules use progressive disclosure. You can skip them when you only need a paper library, reader, and notes.
|
Project library — one evidence space for each research question. |
Project overview — the next action, recent activity, and evidence gaps without a crowded dashboard. |
|
Evidence workflow — reveal screening, extraction, and quality assessment only when needed. |
Reader workspace — source text, annotations, and structured literature notes stay together. |
The screenshots use a synthetic demonstration PDF and public bibliographic metadata. They contain no private user projects or research data. The current UI is Chinese-first.
Install DeepSeek Harness first, then pin the plugin to the published tag:
dsh plugin --profile web add github:zhoupengyun572-cell/dsh-hana-research#v0.5.0-beta.1
Restart Harness after installation. The packaged dsh.bundle activates Hana Research in the web profile without manual configuration edits.
The same package is available from the GitHub Release. SHA-256:
437495d420fdff5268d8fb5431a06b8f7afcb7702ae4aa7133531f6bd9b2a8e7
Update or remove it with:
dsh plugin --profile web update dsh-hana-research
dsh plugin --profile web remove dsh-hana-research
Removing the package does not delete the research database.
$DSH_HOME/plugin-data/hana-research/: SQLite research.db, PDFs, translations, and project-note files.>=22.13.0; Harness baseline @deepseek-ai/dsh-tools 0.1.0-rc.13, with public CI against 0.1.1-rc.2.npm ci
node scripts/run-tests.mjs
cd web
npm ci
node --test tests/*.test.mjs
npm run build
The current baseline is 167/167 root tests and 13/13 reader-editor tests. CI also verifies the release-package allowlist and size budget.
Please include the Harness and Node versions, operating system, reproduction steps, and sanitized logs when reporting a problem.
MIT © 2026 Hana Research contributors