Plugin directory / UI & Skins / dsh-wq-bridge
dsh-wq-bridge
Verified · install-tested on dsh jonah791
What it does
DSH self-built plugin
Works — verified, early-stage project
DSH self-built plugin It installs cleanly and boots without issues in our testing. It's early-stage but functional.
“Verified” means our automated CI actually ran dsh plugin add in a clean profile and it booted — nothing more. Feature descriptions and version compatibility are the author’s claims. This is not a security audit and not an endorsement of third-party code.
README
dsh-wq-bridge — WorldQuant BRAIN 平台桥(Python Bridge 模式)
为 DeepSeek Harness (DSH) 打造的 AlphaFactory 平台桥:TS 壳管理 Python 子进程(stdio JSON-RPC),向 DSH 工具面暴露 AlphaFactory 的平台能力。
状态:v0.3(阶段 1-4 完成:27 工具 + 全主 agent 挖掘循环实证)。DSH 为预览版(0.1.0-rc),无兼容承诺。
安装
cd <你的 self-plugins 目录>
git clone https://github.com/jonah791/dsh-wq-bridge.git
cd dsh-wq-bridge
pnpm install
pnpm build
架构
DSH 工具面 (TS) ←→ PythonBridge (stdio JSON-RPC) ←→ bridge_server.py ←→ AlphaOrchestrator (Python)
- Python 核心 100% 保留(AlphaFactory 215 测试资产不重写)
- 桥为长驻子进程(无每次启动开销);请求-响应按 id 配对,支持并发
- 凭据:读 AlphaFactory 的 .env(WQ_USERNAME/WQ_PASSWORD)注入子进程
- dispose 时 kill 子进程(插件生命周期绑定)
工具面(v0.3 全量 27 工具,按域分组)
平台/只读(5)
| 工具 | 职责 |
|---|---|
wq_ping |
桥连通性(Python 版本) |
wq_alpha_count |
五库计数 |
wq_quota |
WQ 配额监控(挖掘前必查) |
wq_context_explore |
探索上下文(开局必读) |
wq_context_refine |
精炼上下文 |
核心挖掘(12)
| 工具 | 职责 |
|---|---|
wq_simulate |
WQ 回测(~90-150s,300s 桥超时) |
wq_recover_pending |
崩溃兜底(超时后按表达式找回) |
wq_check_expression |
表达式静态校验 |
wq_similarity_hint |
撞车预检(simulate 前必查) |
wq_evaluate_submittability |
四关评估 |
wq_add_to_zoo |
入库(四关过→ready,否则→candidates) |
wq_suggest_next_round |
轮次调度建议 |
wq_report_blindspot |
盲点登记 |
wq_mark_as_waste |
废渣判定 |
wq_add_weak_signal |
弱信号入库(组合素材) |
wq_corr_with / wq_compute_correlation |
PnL 相关实测 |
分析(4,tools.py 纯计算)
| 工具 | 职责 |
|---|---|
wq_analyze_parse |
表达式结构树 |
wq_analyze_similarity |
结构相似度 |
wq_analyze_calibrated |
标定相似度(→PnL corr 估计) |
wq_analyze_classify |
信号族分类 |
知识(6,档案确定性查询)
| 工具 | 职责 |
|---|---|
wq_knowledge_blindspots |
盲点档案(过滤) |
wq_knowledge_summary |
盲点归纳摘要 |
wq_knowledge_meme |
模因库 |
wq_knowledge_templates |
模板库 |
wq_knowledge_family_tree |
血统树 |
wq_knowledge_dead_roots |
判死词根 |
库读取(3)
| 工具 | 职责 |
|---|---|
wq_submitted / wq_ready / wq_weak_pool |
三库读取 |
组合
- insert:
- id: wq-bridge
name: dsh-wq-bridge
测试
node node_modules/typescript/lib/tsc.js -p tsconfig.json && node --test
6/6 tests(mock python 服务:往返/参数/错误/并发/超时/dispose)。
实测证据(2026-08-15)
- 全主 agent 挖掘闭环:quota → context_explore(fn_mne_a 必答格点)→ check/similarity 预检(零撞车)→ simulate(sharpe 0.11,弱信号)→ add_to_zoo(candidates)
- 桥超时教训:simulate 需 300s(WQ 90-150s 波动);recover_pending 兜底找回
- suggest_next_round 真实建议(combo:弱池 4 素材,rule2_combo)
演进
- 后续:四域拆分(mine/analyze/knowledge 独立插件)按需进行;领域知识桥接 dsh-agent-memory
相关
License
MIT
Install
Install the catalog once, then DeepSeek Harness can find and install any plugin from this site automatically:
dsh plugin add dshbase-catalog Then say "install dsh-wq-bridge for me" — your agent finds it in the directory and installs it. Docs: dshbase-catalog · verified packs.
This plugin is GitHub source (not published to npm) — install it straight from the repo:
Web profile:
dsh plugin --profile web add github:jonah791/dsh-wq-bridge Headless (CLI) profile:
dsh plugin --profile headless add github:jonah791/dsh-wq-bridge Test report
Verified: L1 install + L2 load + L3 runtime from GitHub source on dsh 0.1.0-rc.6.