pengpengyi92/dsh-quant
Quantitative R&D toolkit for DeepSeek Harness — 46 tools across six domains covering market data, indicators, factor evaluation, walk-forward ML validation, risk, options, bonds and fund simulation, with an end-to-end research pipeline.
dsh-quant is an AI-native, DSH-native quant toolkit: 59 quant_* tools across six pluggable domains — data (market data/sources/quality), alpha (indicators/factors/eval), model/ML (backtests, ML/DL/RL frameworks), risk (VaR/drawdown/options/bonds), execution (sim execution/fund/report) and ecosystem (GitHub/npm stats, influence pulse) — assembled as five pluggable modules plus a community layer, with an end-to-end PDAT→PET research pipeline (data → quality → factors → backtest → paper trading) that runs on public BTC data in about three minutes. The design is deliberately agent-first: tool schemas are injected into the system prompt, outputs use equal-length null alignment (leading window positions are null so the model aligns by index), canonical JSON is separated from human-readable rendering, all 59 tools are isConcurrencySafe pure functions with no shared state, and a skill/quant-research layer lets the model load the workflow itself. 215 unit tests, zero runtime deps, offline-verifiable; data and conclusions stay internal while tools and methods ship open.
Install
npm i dsh-quantnpm dsh-quant 0.90.0 verified 2026-09-04 (repository field empty — npm maintainer pengyipeng_1, email pengpengyi92@gmail.com, matches repo owner pengpengyi92, owner-name-consistent). README quick install for dsh users: npm i dsh-quant, then add one line to cordis.yml: - name: 'dsh-quant' — 59 quant_* tools auto-register (indicators / backtests / factors / risk / fund simulation / ecosystem metrics). Zero runtime dependencies; 21 TypeScript source files, no Python. Site: dsh-quant-site.pages.dev.
Compatibility
DeepSeek Harness plugin running in the harness's Node runtime (same process as the agent, composable by the Loader, HMR-safe reversible registration); zero runtime dependencies; pure-function tools with hand-computed baselines; listed in awesome-dsh-plugin.
Details
- Repo: pengpengyi92/dsh-quant
- Category: Tools & Capabilities
- Stars: 0
- Version: npm dsh-quant 0.90.0
- Last push: 2026-09-18
- First seen: 2026-08-16
Recent updates
0.90.0 is the current npm latest (verified 2026-09-04).
FAQ
- How is it installed?
- Per the README quick install: npm i dsh-quant, then add - name: 'dsh-quant' to your cordis.yml — 59 quant_* tools auto-register on the next profile start.
- What can the agent actually do with it?
- The full open-source PDAT→PET quant loop: fetch market data, run quality checks, evaluate factors, backtest a strategy, and run a paper-trading simulation — the README demo goes from public BTC data to a paper pipeline in ~3 minutes.
- Does it need Python or external services?
- No — zero runtime dependencies, 21 TypeScript source files, no Python; the framework's pure-function tools are offline-verifiable with 215 unit tests.
Alternatives
linxichen/dsh-rigorquant · AllenCX/dsh-quant-workspace