Local-first A-share research workbench on DeepSeek Harness: market dashboards, watchlist valuation, K-line turning-point markers, master-method research reports and open expert chats, with versioned archives and follow-up on the original DSH session.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-08-27
License
MIT
01What can it help you accomplish?
Run a complete local-first A-share research loop inside DeepSeek Harness
Market overview, watchlist valuation, per-stock quotes with K-line observation, master-method research, open expert conversations, report archiving and continuous follow-up in one UI
A-share investors and researchers who run DeepSeek Harness (dsh) locally and want an agent-driven workbench from discovery to continuous validation
Produce archivable master-method research reports and keep the conversation going after the report
Verifiable, immutable Markdown reports with versions and hashes from master-method research, plus continuous follow-up on the original dshSessionId with tool-process visibility
A-share researchers who want DSH agent research persisted as traceable documents and continuously updated
02How to install into DeepSeek Harness
Prerequisites
- Node.js `^22.19.0` or `>=24.0.0` and pnpm `11.7.0`
- DeepSeek Harness `0.1.1-rc.2` (DSH is still pre-release)
- A DeepSeek API Key (only needed when actually running agents)
Installation steps
- 01
Clone the repo: `git clone git@github.com:hanai-labs/worth-dsh.git`
$ git clone git@github.com:hanai-labs/worth-dsh.git
- 02
Enter the directory: `cd worth-dsh`
- 03
Install dependencies: `pnpm install`
- 04
Build: `pnpm run build`
- 05
Install the profile: `pnpm run profile:install -- --package .` (for the published package, use `--package hanai-investment-dsh` instead)
- 06
Verify the assembly: `pnpm run profile:verify`
- 07
Launch: `dsh --profile hanai-investment`
$ dsh --profile hanai-investment
Verify the integration
- Run `pnpm run profile:verify` to verify the profile assembly
- Run `pnpm run check` to execute the full gate suite
03DSH integration and capability boundaries
Installed as a dedicated `hanai-investment` Profile and launched natively through the official dsh CLI (`dsh --profile hanai-investment`); the product UI runs on DSH AppFrame's full-screen shell.overlay plugin canvas, with DSH as the single source of truth for models, agents, tools, sessions and streaming events
Market overview & watchlist valuation
Six major indices, sector/concept treemap and four ranking lists; watchlist groups and stocks→Market snapshot plus watchlist quotes and fair-value estimates (PE/PB, amount and ratio vs current price), with fresh/stale/unavailable states
Multi-timeframe K-line & turning-point markers
Forward-adjusted intraday / daily / weekly / monthly K-line with MA5/10 and MA20/60→Ten turning-point marker types with historical-evidence tooltips, dynamic-vs-closed calculation states and a show/hide switch
Master-method research & open expert chats
A stock plus a master, or an expert chosen directly without selecting a stock→Verifiable Markdown reports from four master methodologies; persistent open conversations with five expert perspectives (the Sun Yuchen persona discloses its AI simulation boundary)
Each research run and chat binds its own persistent DSH SessionVersioned reports & post-report chat
Completed or failed master research reports→Immutable report versions with hashes and file sizes; follow-up on the original dshSessionId with queued send, steering, cancellation and approvals
Deletion is restricted to finished/failed reports whose session is not runningFull-market backtest & validation of turning points
5,813 current and historical A-share securities; 9,813,872 valid tradable daily K-lines (2014-01-02 to 2026-08-20)→Frozen V0-rule retrospective direction frequencies, matched controls and Holm-corrected conclusions; 0/16 rules passed the strict release gate
04Who is it for? When not to use it?
Good for
- A-share investors and researchers who run DeepSeek Harness (dsh) locally and want an agent-driven workbench from discovery to continuous validation
- A-share researchers who want DSH agent research persisted as traceable documents and continuously updated
Not for
- DeepSeek Harness 0.1.1-rc.2 is required and DSH is still pre-release: rc versions are not guaranteed compatible, so re-verify before upgrading, and keep the CLI, Web App and Hanai on the same version.
05Compatibility, maintenance and safety notes
- DeepSeek Harness 0.1.1-rc.2 is required and DSH is still pre-release: rc versions are not guaranteed compatible, so re-verify before upgrading, and keep the CLI, Web App and Hanai on the same version.
- Requires Node.js ^22.19.0 or >=24.0.0 and pnpm 11.7.0; a DeepSeek API Key is only needed when actually running agents.
- Quotes rely on public sources such as Eastmoney and Tencent and degrade to delayed feeds when the realtime cluster is unavailable; the GuruFocus interface is a personal-research prototype only, and production or commercial deployment should switch to a data provider with formal authorization and SLA.
- The project is a research aid, not investment advice; it makes no promise of data realtime-ness, completeness or returns, and turning-point markers are volume-price observation hints filtered by historical condition frequency, not buy/sell points.
- DSH session deletion and cross-version migration are still limited; backing up an ongoing conversation requires keeping both the Hanai data root and the DSH session data.
MIT · actively maintained (no tagged release; last push 2026-08-26)
06Frequently asked questions
How do I install hanai-investment-dsh?
From source: `git clone git@github.com:hanai-labs/worth-dsh.git`, then `pnpm install`, `pnpm run build`, `pnpm run profile:install -- --package .`, `pnpm run profile:verify`, and launch with `dsh --profile hanai-investment`. To install the published package instead, replace `--package .` with `--package hanai-investment-dsh`; re-running the installer is safe.
How does it relate to DeepSeek Harness?
DSH is the single source of truth for models, agents, tools, sessions and streaming events; Hanai keeps no second messages/turns table — its SQLite only stores watchlists, security master data and opaque dshSessionIds. The UI is built with React 18, DSH Slot/Runtime and ECharts, and does not reuse DSH's native chat UI.
What are the runtime requirements?
Node.js ^22.19.0 or >=24.0.0, pnpm 11.7.0, DeepSeek Harness 0.1.1-rc.2, and a DeepSeek API Key (only needed when actually running agents). DSH is still pre-release, so re-verify before upgrading to another rc and keep the CLI, Web App and Hanai on the same version.
Where is data stored, and does it go online?
Hanai business data defaults to ~/.hanai-investment-dsh/ with 0700/0600 permissions, and the API key is a write-only secret. Quotes come from public sources such as Eastmoney and Tencent, degrading to delayed feeds when the realtime cluster is unavailable; the browser never touches the filesystem or SQLite — all writes go through the same-origin /hanai RPC.
Are the turning-point markers buy/sell advice?
No. Turning points are volume-price observation hints filtered by historical condition frequency; online triggering calls no large model and only runs deterministic computation on already-fetched forward-adjusted OHLCV. The project is a research aid and provides no investment advice, realtime/data-completeness or return promises.
07Related DSH workflows
dsh-web
by zhu1090093659
One-command install of a full DSH web workbench: monitoring HUD, task board, SSH panel, mobile remote, vision tool, git graph, skins, plugin manager.
petdex
by crafter-station
A public gallery of animated pets for Codex, Claude Code, DeepSeek Harness, Hermes, OpenCode, Gemini CLI, and more.
j-space-cognition-suite-v3.7
by tiger3807861189
AI cognitive-enhancement skills for agent harnesses inspired by Anthropic's global workspace research. Adds inference-time control and reasoning tools for DeepSeek Harness, Claude Code, Codex and OpenCode.
dsh-deep-whale
by small-tailqwq
DSH Web GUI 鲸鱼娘主题皮肤（深海女仆工坊 maid-atelier：亮/暗双模式）
08Data and sources
DeepSeek Harness（DSH）负责模型、Agent、工具、Session、流式事件和会话持久化；Hanai Worth 负责证券与估值数据、自选分组、研究资料、不可变报告快照和全部产品界面。
安装器会创建或安全迁移独立的 `hanai-investment` Profile。最终 Bundle 顺序固定为 DSH Base、DSH Web App、Hanai；只有 `hanai-investment-dsh` 是 Profile…
`shell.overlay` 是 DSH AppFrame 的全屏插件画布，不是弹窗、iframe 或命令行 Shell。
使用 React 18、DSH Slot/Runtime、ECharts 和 CSS Modules；不显示或复用 DSH 原生聊天 UI。
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-08-27. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.