Quantmind Operations logo

Quantmind Operations

CommunityPopular
qusong0627
quantmind-operations

QuantMind 平台运营操作技能 — 覆盖模型训练、模型管理、后台数据更新、字段信息查询、RSS 新闻对接与分析。在 QuantBot / Claude Code 中处理模型训练、数据同步、新闻分析等任务时使用。触发词:模型管理、数据更新、字段信息、RSS、新闻分析、查看数据、同步数据

Overview

Publisherqusong0627
RepositoryQuantMind
Skill namequantmind-operations
Stars
1.7K
Forks
387
Bundled files
2
LicenseAGPL-3.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by qusong0627 on GitHub. Read the source before you install it.

Installation

Install the Quantmind Operations AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/qusong0627/QuantMind/tree/master/skills/quantmind-operations
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/qusong0627/QuantMind.git /tmp/QuantMind
mkdir -p .claude/skills
cp -r /tmp/QuantMind/skills/quantmind-operations .claude/skills/quantmind-operations
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Quantmind Operations in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Quantmind Operations on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Quantmind Operations is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

⚙️ 本技能遵循公共运行环境契约(最高优先级,先于本文其余内容执行): 详见 _shared/env-contract.md,执行前先读它。

QuantMind 运营操作技能

QuantMind 量化平台的完整运营操作指南。所有 API 都通过 API 网关(默认 http://127.0.0.1:8000 或 http://192.168.31.68:3080)访问,统一加 /api/v1 前缀。

认证

所有请求需要 Bearer Token:

bash
# 获取 token(admin 账号)
TOKEN=$(curl -s -X POST $BASE/api/v1/auth/login \
  -H "Content-Type: application/json" \
  -d '{"username":"admin","password":"admin123","tenant_id":"default"}' \
  | python3 -c "import sys,json; print(json.load(sys.stdin).get('access_token',''))")

# 通用请求头
AUTH="Authorization: Bearer $TOKEN"
CT="Content-Type: application/json"

1. 模型训练(5 步流程)

模型训练分 5 步,与前端 ModelTrainingPage 一致:

特征选择 → 训练目标 → 参数配置 → 执行训练 → 结果入库

1.1 特征选择(筛选输入因子)

bash
# 获取特征字典(类别/数量由 QuantDB l1_factors 动态生成,以接口返回为准)
curl -s -H "$AUTH" "$BASE/api/v1/models/feature-catalog"

# 带数据覆盖统计(含建议训练/验证/测试区间)
curl -s -H "$AUTH" "$BASE/api/v1/models/feature-catalog?include_coverage=true"

# 管理端特征字典(含扫描详情)
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/feature-catalog"

选择特征 key 列表(如 ["mom_ret_5d", "vol_std_20"])或按类别(feature_categories)。 特征类别由后端特征目录动态生成,随 QuantDB l1_factors 数据版本变化(示例 version 20260831 返回 10 类 110 特征:momentum / fundamental / money_flow / style / technical / turnover / concept / volatility / chip / industry)。先读接口返回的 categories[].id,不要硬编码类别清单。

1.2 训练目标(定义 T+N 标签口径)

  • target_horizon_days:预测周期(T+1 / T+5 / T+20 等)
  • target_mode:regression(回归)或 classification(分类)
  • label_formula:标签计算公式(如 close_future/close - 1)
  • effective_trade_date:生效交易日
  • training_window:训练窗口(如 rolling)

1.3 参数配置(设置超参与训练上下文)

  • 时间划分:train_start/end、valid_start/end、test_start/end、val_ratio
  • 模型超参:num_boost_round、early_stopping_rounds、lgb_params/xgb_params/catboost_params/dl_params
  • 训练上下文 context:initial_capital、benchmark、commission_rate、slippage、deal_price、market、industry_as_feature

1.4 执行训练(编排请求与日志预览)

bash
curl -s -X POST "$BASE/api/v1/models/run-training" -H "$AUTH" -H "$CT" -d '{
  "model_type": "lightgbm",
  "model_types": ["lightgbm", "xgboost", "catboost"],
  "ensemble": "stacking",
  "job_name": "我的模型",
  "display_name": "我的模型",
  "train_start": "2022-01-01",
  "train_end": "2024-12-31",
  "valid_start": "2023-06-01",
  "valid_end": "2024-06-30",
  "test_start": "2024-07-01",
  "test_end": "2024-12-31",
  "val_ratio": 0.15,
  "num_boost_round": 1000,
  "early_stopping_rounds": 100,
  "features": ["mom_ret_5d", "vol_std_20"],
  "feature_categories": ["momentum", "volatility"],
  "target_horizon_days": 1,
  "target_mode": "regression",
  "label_formula": "close_future/close - 1",
  "effective_trade_date": "2025-01-02",
  "training_window": "rolling",
  "context": {"initial_capital": 1000000, "benchmark": "000300.SH", "commission_rate": 0.0003, "slippage": 0.001, "deal_price": "close", "market": "CN", "industry_as_feature": false},
  "deploy_to_production": false
}'

支持的 model_type(13 种,以 backend/shared/training/request.py::ALLOWED_MODEL_TYPES 为准):

  • 树/线性:lightgbm / xgboost / catboost / linear / random_forest
  • 深度学习:gru / lstm / alstm / transformer / tabnet / tcn / nativetft
  • 其他:mlp(sklearn 实现)
  • ⚠️ hybrid_gru_tree 已剔除(QLIB map 无实现),勿再传(会被 422 拒绝/落入不支持)。

ensemble 取值:none / stacking / blending / voting(多模型训练时生效) 可选高级参数:wfa(walk-forward,rolling/expanding)、target_horizon_days(单周期 T+N,1–30)、各模型专属超参 lgb_params/xgb_params/catboost_params/dl_params

⚠️ 已下线/死配置:horizons(多周期,2026-09 随多周期训练一并清理)、optuna、n_folds(只建类型不参与序列化,传了不生效)。 返回:runId + 有效/缺失特征统计

1.5 结果入库(查看元数据与产物)

bash
# 轮询训练状态(pending→running→completed/failed)
curl -s -H "$AUTH" "$BASE/api/v1/models/training-runs/{run_id}"

# 训练完成后模型进入 /models,可设为默认
curl -s -X PATCH -H "$AUTH" -H "$CT" "$BASE/api/v1/models/default" -d '{"model_id":"xxx"}'

# 查看模型列表确认入库
curl -s -H "$AUTH" "$BASE/api/v1/models"
curl -s -H "$AUTH" "$BASE/api/v1/models?include_archived=true"

# 系统内置模型
curl -s -H "$AUTH" "$BASE/api/v1/models/system-models"

# ⚠️ 手工融合模型已下线:原 ensemble 创建路由不存在
# (多周期 + 手工融合已于 2026-09 清理,见 backend/scripts/cleanup_multi_horizon_ensemble.py;
#   仅在多模型训练时用 model_types + ensemble 合成,或保留历史融合模型做推理兼容)

2. 模型管理(管理端)

2.1 扫描本地模型目录

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/scan"

2.2 数据状态

bash
# Qlib + 特征快照数据状态
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/data-status"

2.3 推理前置检查(生成明日信号)

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/precheck-inference"

2.4 滚动回测

bash
curl -s -X POST "$BASE/api/v1/admin/models/backtest" -H "$AUTH" -H "$CT" -d '{
  "model_id": "xxx",
  "start": "2024-01-01",
  "end": "2024-12-31"
}'
# 可用回测日期
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/backtest/trading-dates"
# 回测历史
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/backtest/history/{model_id}"

2.5 推理回测(选股策略事件驱动)

bash
curl -s -X POST "$BASE/api/v1/admin/models/inference-backtest" -H "$AUTH" -H "$CT" -d '{
  "model_id": "xxx"
}'

3. 后台数据更新(五市场)

3.1 统一日同步(推荐)

bash
# 提交同步任务,返回 task_id(market: A/CN=QuantDB, US=QuantUS, HK=QuantHK, BC=区块链, FUTURES=期货)
curl -s -X POST "$BASE/api/v1/admin/data-platform/daily-sync" -H "$AUTH" -H "$CT" -d '{
  "market": "A",
  "symbols": [],
  "incremental": true,
  "calibrate": true
}'
# 查询同步状态
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/daily-sync/status/{task_id}"

各市场同步数据源:

市场market 值数据源说明
A股A / CNQuantDB SDK4阶段:parquet→PG→Qlib→特征快照
美股USYahoo Financequantus_daily_sync.py
港股HKYahoo + akshare + CCASSquanthk_daily_sync.py
区块链BCBinancequantbc_daily_sync.py(支持 --minute)
期货FUTURESaksharequantfutures_daily_sync.py

3.2 定时同步调度(每市场独立配置)

bash
# 查看全部市场定时配置
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/sync-schedule"
# 查看单市场配置
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/sync-schedule/{market}"
# 保存单市场配置(enabled/time/days/datasets/with_qlib)
curl -s -X POST -H "$AUTH" -H "$CT" "$BASE/api/v1/admin/data-platform/sync-schedule/{market}" \
  -d '{"enabled": true, "time": "22:30", "days": [1,2,3,4,5], "datasets": ["all"], "with_qlib": true}'
# 立即触发一次同步(测试)
curl -s -X POST -H "$AUTH" "$BASE/api/v1/admin/data-platform/sync-schedule/{market}/run"

3.3 同步状态 / 进度

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/sync-status"
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/sync-progress"

3.4 Qlib 同步(增量重建缓存)

bash
# 同步数据集时带 with_qlib 触发 Qlib 缓存重建
curl -s -X POST -H "$AUTH" -H "$CT" "$BASE/api/v1/admin/data-platform/quantdb/sync-datasets" \
  -d '{"datasets":["l1_factors","l2_factors"],"with_qlib":true}'
# 查看 Qlib + 特征快照数据状态(含年度快照详情)
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/data-status"

Qlib 路径(QlibDataBuilder.for_market):A股 .qlib_cache/cn_data,HK/US/BC/FUTURES 各目录下 .qlib_cache/{hk,us,bc,futures}_data。

3.5 特征快照(更新特征 parquet)

bash
# 指定年份(A股按年生成 model_features_{year}.parquet)
curl -s -X POST -H "$AUTH" "$BASE/api/v1/admin/data-platform/update-feature-parquet?year=2026"
# 多市场特征更新(非 A 股市场,market 必填:hong_kong / us_stock / crypto)
curl -s -X POST -H "$AUTH" "$BASE/api/v1/admin/data/update-market-features?market=hong_kong"
# 特征快照年度详情(A股逐年 metadata.json)
curl -s -H "$AUTH" "$BASE/api/v1/admin/models/data-status"

特征快照结构:A股 db/feature_snapshots/model_features_{year}.parquet(含 .metadata.json 年度详情),非A股单体 model_features_{market}.parquet。

3.6 基本面同步 / 数据新鲜度

bash
curl -s -X POST -H "$AUTH" "$BASE/api/v1/admin/data-platform/sync-fundamentals"
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/freshness"

3.7 在线状态 / 数据源健康

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/online-status"
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/sources"
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/sources/{name}/health"

3.8 万得 L2 原始数据导入(手动,不走 daily-sync)

逐笔委托/成交/十档盘口由 wind_l2_import.py 从万得逐日 7z 手动导入 (schema/单位/坑见 [[quantdb-fields]] 第三章,含深市成交量≈2×、tick_data 单位混源):

bash
python backend/scripts/wind_l2_import.py --archive /path/to/20260511.7z                  # 全市场
python backend/scripts/wind_l2_import.py --archive /path/to/20260511.7z --symbols 000001.SZ

落盘 1_kline_data/l2_data/order_|trade_{code}_{date}.parquet + tick_data/{code}_{date}.parquet; 文件名即日期(20260511.7z → 20260511),增量跳过已存在,--force 覆盖,可断点续跑。

4. 字段信息

4.1 字段覆盖矩阵(市场 × 字段 × 源)

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/health-matrix?market=A"

4.2 字段覆盖表

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/field-coverage"

4.3 质量告警

bash
curl -s -H "$AUTH" "$BASE/api/v1/admin/data-platform/quality-alerts"

4.4 支持的字段类别(特征字典)

通过 /api/v1/models/feature-catalog 获取。类别与特征数由 QuantDB l1_factors 动态生成(示例 version 20260831 返回 10 类 110 特征),随数据版本变化——以接口返回为准,不要硬编码类别清单。

6. 推理研究(推理中心 + 推理历史)

推理研究涵盖:单日推理、批量多日推理、批量单日推理、推理历史、股票历史分数。

6.1 推理前置检查

bash
# 生成明日信号前置检查(确认数据就绪)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/precheck"

6.2 单日推理(核心)

bash
# 对指定模型在指定日期执行推理(可能耗时数分钟)
curl -s -X POST "$BASE/api/v1/models/inference/run" -H "$AUTH" -H "$CT" \
  -d '{"model_id":"xxx", "inference_date":"2026-08-07"}' \
  -w "\nHTTP %{http_code}\n"

6.3 批量推理(单日批量 / 多日批量)

批量推理支持两种模式,提交后立即返回 batch_id,逐日推理在后台执行:

A. 批量单日推理(range 模式)——区间内每个交易日逐日执行单日推理

bash
curl -s -X POST "$BASE/api/v1/models/inference/batch" -H "$AUTH" -H "$CT" -d '{
  "model_id": "xxx",
  "mode": "range",
  "start_date": "2026-08-01",
  "end_date": "2026-08-07",
  "top_k": 20,
  "side": "both"
}'

B. 批量多日推理(lookback 模式)——锚定日回溯 N 个交易日

bash
curl -s -X POST "$BASE/api/v1/models/inference/batch" -H "$AUTH" -H "$CT" -d '{
  "model_id": "xxx",
  "mode": "lookback",
  "anchor_date": "2026-08-07",
  "window_days": 30,        # 默认 = 模型 horizon,所有信号梯队仍持有中
  "top_k": 20,
  "side": "both",
  "reuse_existing": true
}'

参数完整说明:

参数取值说明
moderange / lookbackrange=日期区间逐日;lookback=锚定日回溯窗口
start_date / end_dateYYYY-MM-DDrange 模式必填,区间内逐日推理
anchor_dateYYYY-MM-DDlookback 模式必填
window_days整数lookback 回溯天数(默认=模型 horizon)
top_k整数每日排名前 N 名
sidelong / short / both多/空/双向
reuse_existing布尔复用已存在的推理结果
concurrency整数并发度

返回:HTTP 202 + batch_id。之后用 batch_id 轮询进度。

6.4 批量推理历史与进度

bash
# 批量推理历史
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/batches?page=1&page_size=20"
# 单个批次进度(status: pending/running/completed/failed)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/batch/{batch_id}"
# 删除批次记录
curl -s -X DELETE -H "$AUTH" "$BASE/api/v1/models/inference/batch/{batch_id}"

6.5 批量推理实战流程

  1. 确认模型:/models/default 或 /models 选模型
  2. 提交:range(指定区间)或 lookback(锚定+窗口)
  3. 轮询:/inference/batch/{batch_id} 查进度,completed 后取结果
  4. 汇总:批量结果含每日信号,可对比多日信号变化
  5. 清理:不需要的批次 DELETE

6.6 推理历史(单日推理记录)

bash
# 推理历史(支持按 run_id/状态/日期过滤)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/runs?model_id=xxx&page=1&page_size=20"
# 单次推理结果明细(排名/信号/行业等)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/runs/{run_id}"
# 删除推理记录
curl -s -X DELETE -H "$AUTH" "$BASE/api/v1/models/inference/runs/{run_id}"

6.7 单只股票历史推理分数

bash
# 某股票的历史推理分数趋势(用于交叉验证选股)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/stock/600036.SH/history?days=180"

6.8 推理自动设置 / 最新批次

bash
# 自动推理设置(每日定时)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/settings/{model_id}"
curl -s -X PUT -H "$AUTH" -H "$CT" "$BASE/api/v1/models/inference/settings/{model_id}" -d '{"auto_enabled": true}'
# 当前生效推理批次
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/latest"

6.9 批量聚合分析(推理分析)

bash
# 某批次的聚合分析(per_symbol/groups/movers/daily/meta,含 IC/趋势/共识带)
curl -s -H "$AUTH" "$BASE/api/v1/models/inference/batch/{batch_id}/aggregate"

6.10 融合模型 pred 生成(回测信号)

融合模型(ensemble_config.json)本身无 pred.pkl,AI-IDE 回测/信号生成时会自动调用 generate_ensemble_pred:读取子模型 pred.pkl → 按 (datetime, instrument) 对齐 → 截面排名百分位加权融合 → 落到融合模型目录 pred.pkl。单模型无 pred 时提示"请先推理"。

7. RSS 新闻对接与分析

5.1 新闻源列表

bash
curl -s -H "$AUTH" "$BASE/api/v1/news/sources"
# 返回: {sources: [{source_id, source_name, subscribe_url, type, folder_id, folder_name}], folders, total}

5.2 拉取新闻文章(核心接口,支持丰富过滤)

bash
curl -s -H "$AUTH" "$BASE/api/v1/news/articles" \
  -G \
  --data-urlencode "tickers=600519.SH,000858.SZ" \
  --data-urlencode "industries=白酒,消费" \
  --data-urlencode "sentiment=bullish" \
  --data-urlencode "event_tags=财报,业绩预增" \
  --data-urlencode "keyword=茅台" \
  --data-urlencode "sort=sentiment_bullish" \
  --data-urlencode "since=2026-08-01T00:00:00Z" \
  --data-urlencode "page=1"

过滤参数:

  • source_id / source_ids — 新闻源过滤
  • folder_id — 文件夹过滤
  • keyword — 标题关键词
  • tickers — 股票代码(逗号分隔)
  • industries — 行业
  • sentiment — bullish / bearish / neutral
  • event_tags — 事件标签(财报/业绩预增/减持等)
  • countries / regions — 国家/地区
  • key_terms — 关键词(AI/半导体等)
  • date_entities — 提及日期
  • starred — 仅收藏
  • strong_only — 仅强信号(|score|>=0.5)
  • sort — time_desc(最新)/ time_asc / sentiment_bullish(利好强度)/ sentiment_bearish(利空强度)

5.3 单篇文章详情

bash
curl -s -H "$AUTH" "$BASE/api/v1/news/articles/{article_id}"

5.4 新闻富化统计 / 触发富化

bash
curl -s -H "$AUTH" "$BASE/api/v1/news/enrichment/stats"
curl -s -X POST -H "$AUTH" "$BASE/api/v1/news/enrichment/run"
curl -s -X POST -H "$AUTH" "$BASE/api/v1/news/enrichment/rebuild-all"

5.5 刷新新闻源

bash
curl -s -X POST -H "$AUTH" "$BASE/api/v1/news/sources/{source_id}/refresh"

8. 实战分析流程(推荐顺序)

当用户要求分析某股票/行业时,按此流程:

  1. 查新闻:/news/articles 带 tickers + sentiment + since,看利好/利空
  2. 查模型分数:/models/inference/stock/{symbol}/history 看历史推理分数趋势
  3. 查数据健康:/admin/data-platform/health-matrix?market=A 确认数据完整
  4. 查当前模型:/models/default 确认生效模型
  5. 需要更新数据:/admin/data-platform/daily-sync 提交增量同步
  6. 需要训练:先 feature-catalog 拿字段,再 run-training

当用户要求挖掘新因子时,使用 [[rd-agent-factor-mining]] 技能(RD-Agent 自动演化管线)。 当用户要求按条件选股 / 筛选股票池时,使用 [[smart-strategy-stock-picking]] 技能(基于 QuantDB 字段字典的条件选股)。 当用户要求查询 QuantDB 数据 / 配置 API Key / 查看数据集字段时,使用 [[quantdb-sdk]] 技能。 当用户要求深度分析市场 / 数据挖掘 / 导出分析数据 / 生成投研报告时,使用 [[stock-market-analysis]] 技能。 当用户要求运行回测 / 对比策略 / 参数优化 / 分析回测结果时,使用 [[backtest-center]] 技能。 当用户要求用 AI 写策略 / 生成 Qlib 策略代码时,使用 [[ai-ide-strategy-writing]] 技能。 当用户要求模拟交易 / 下单 / 查持仓时,使用 [[simulation-trading]] 技能。 当用户要求分析批量推理结果 / 解读信号 / 选股决策 / 负分参考时,使用 [[batch-inference-analysis]] 技能。 当用户要求生成投研报告 / 深度研报 / 多Agent分析时,使用 [[stock-deep-research]] 技能。

注:投研报告由 [[stock-deep-research]] 技能(智能体自主版,任意大模型可跑)编排生成,落盘后由「技能中心 → 报告档案」统一浏览。

9. 相关技能

  • [[rd-agent-factor-mining]] — 自动调用 RD-Agent 挖掘因子(evolve/tasks/factors/backtest/export)
  • [[smart-strategy-stock-picking]] — 基于 QuantDB 数据的条件选股(自然语言/条件/DSL 三种方式)
  • [[quantdb-sdk]] — QuantDB 数据 SDK(API Key 配置、28 数据集目录、字段查询、远程查询、同步)
  • [[stock-market-analysis]] — 市场深度分析 + 数据导出(全市场扫描/行业轮动/个股371字段/风险评分/CSV导出)
  • [[backtest-center]] — 回测中心(快速回测/专家模式/策略对比/参数优化/高级分析/向量化极速回测)
  • [[ai-ide-strategy-writing]] — AI-IDE 写策略并执行(Docker runner 运行/回测)
  • [[simulation-trading]] — 模拟交易(下单买卖/持仓/成交/账户/模拟盘启动)
  • [[batch-inference-analysis]] — 批量推理结果分析(市场状态/选股/负分参考/行业轮动)
  • [[stock-deep-research]] — 投研分析(智能体自主版:本地数据 → 多空子代理辩论 → 综合研判 → PDF 报告归档)

10. 常见排查

现象排查
特征字典加载失败/models/feature-catalog 返回是否 200,看服务健康
数据匹配不到/admin/data-platform/health-matrix 看字段覆盖,/freshness 看新鲜度
新闻空白/news/sources 确认源存在,/news/enrichment/stats 看富化状态
训练失败/models/training-runs/{run_id} 查状态,看 features 是否在 parquet 中存在

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Quantmind Operations AI skill do?

QuantMind 平台运营操作技能 — 覆盖模型训练、模型管理、后台数据更新、字段信息查询、RSS 新闻对接与分析。在 QuantBot / Claude Code 中处理模型训练、数据同步、新闻分析等任务时使用。触发词:模型管理、数据更新、字段信息、RSS、新闻分析、查看数据、同步数据

Why use Quantmind Operations on TypingMind?

Because you install it once and use it with any model. Quantmind Operations is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Quantmind Operations in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/qusong0627/QuantMind/tree/master/skills/quantmind-operations. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Quantmind Operations?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Quantmind Operations?

As many as you like. As long as a model supports skills, you can use Quantmind Operations with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Quantmind Operations AI skill free?

Yes. It is published on GitHub by qusong0627 under the AGPL-3.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇