darts-forecasting

Darts 是轻量级时间序列预测库，支持多市场金融数据的确定性与概率性预测，提供协变量整合与层级聚合能力。触发场景：(1) 用户要同时对多个市场的金融数据做时间序列预测；(2) 用户要预测结果带置信区间，评估不确定性范围；(3) 用户要用外部变量（宏观指标、市场情绪）辅助提升预测精度。

Security Scan

The skill's files, install script, and runtime instructions are consistent with a Darts + ZVT time-series/backtest helper — nothing requested is disproportionate to that purpose, though there are a few small metadata and workflow inconsistencies worth noting before installation.

Purpose & Capability

The skill describes a Darts-based multi-market forecasting toolkit integrated with ZVT for A-share/HK/crypto data — the files, use-cases, and references are aligned with that purpose. Minor inconsistencies: SKILL.md metadata claims 'Requires Python 3.12+ with uv package manager' while scripts/install.sh uses python3/pip; the published registry version (0.3.0) differs from SKILL.md/seed.yaml compiled version (v6.1). LICENSE.txt is referenced but not present in the manifest.

Instruction Scope

SKILL.md and seed.yaml instruct the agent to re-read seed.yaml, run preconditions, and validate zvt environment (including read/write checks on ~/.zvt). Those file- and filesystem-related checks are related to ZVT operation and backtests, but the agent will perform filesystem checks and may prompt or ask users to run recorders; the instructions do not request unrelated system secrets or network endpoints beyond expected data providers.

Install Mechanism

No remote arbitrary downloads; scripts/install.sh runs pip installs of common scientific packages (numpy, pandas, scikit-learn, etc.), a standard low-risk install path. No unknown URLs, archives, or extract steps detected.

Credentials

The skill declares no required environment variables or credentials. It references external data providers (eastmoney, joinquant, baostock, akshare) which may require separate user credentials in practice, but the skill itself does not demand unrelated secrets or excessive environment access.

Persistence & Privilege

Flags: always=false (not force-included) and agent invocation is allowed (normal). The skill does include rules that instruct re-reading its seed.yaml and running preconditions, but it does not request to change other skills or system-wide agent settings.

Assessment

What to consider before installing: (1) Run the install script in an isolated Python virtualenv (do not run as root); scripts/install.sh only runs pip installs, but confirm your environment uses the intended Python version (SKILL.md says Python 3.12+ while the script uses the system python3). (2) The skill integrates Darts and ZVT and may check/create ~/.zvt and require running ZVT recorders to fetch market data — expect filesystem writes and data-collection steps related to market recorders. (3) No credentials are required by the skill itself, but some data sources (joinquant, qmt/broker) require accounts — provide those separately and only to trusted code. (4) Metadata mismatches (version numbers, missing LICENSE.txt) are non-critical but worth verifying with the publisher if you need provenance or license clarity. (5) If you will run any generated trading/execution code, review semantic locks (SL-* rules) and test backtests in a sandboxed environment before connecting to live brokers.

I help you build quant strategies on A-share with ZVT — from data fetch to backtest, one flow. Just tell me what you want; I'll write the code, you don't have to dig docs. (Heads up: ZVT natively supports A-share, HK, and crypto. US stocks — stockus_nasdaq_AAPL — are half-baked; don't bother for serious work.)

Pipeline

data_collection -> data_storage -> factor_computation -> target_selection -> trading_execution -> visualization

Top Use Cases (31 total)

Sphinx Package Title Fixer (UC-101)

Automates extraction of descriptive titles and docstrings from Python packages to improve Sphinx API documentation readability Triggers: sphinx documentation, package titles, docstring extraction

Sphinx Documentation Configuration (UC-102)

Configures Sphinx documentation builder with extensions for auto-summary, autodoc, and graphviz visualization Triggers: sphinx config, documentation, autodoc

Example Utilities Module (UC-131)

Provides utility functions for managing Python paths when running Darts examples locally Triggers: utilities, path management, example helpers

For all 31 use cases, see references/USE_CASES.md.

Install

# One-time setup before first use
bash scripts/install.sh

Execute trigger: When user intent matches intent_router.uc_entries[].positive_terms AND user uses action verb (run/execute/跑/执行/backtest/fetch/collect)

What I'll Ask You

• Target market: A-share (default), HK, or crypto? (US stocks in ZVT are half-baked — stockus_nasdaq_AAPL exists but coverage is thin)
• Data source / provider: eastmoney (free, no account), joinquant (account+paid), baostock (free, good history), akshare, or qmt (broker)?
• Strategy type: MACD golden-cross, MA crossover, volume breakout, fundamental screen, or custom factor?
• Time range: start_timestamp and end_timestamp for backtest period
• Target entity IDs: specific stocks (stock_sh_600000) or index components (SZ1000)?

Semantic Locks (Fatal)

ID	Rule	On Violation
SL-01	Execute sell orders before buy orders in every trading cycle	halt
SL-02	Trading signals MUST use next-bar execution (no look-ahead)	halt
SL-03	Entity IDs MUST follow format entity_type_exchange_code	halt
SL-04	DataFrame index MUST be MultiIndex (entity_id, timestamp)	halt
SL-05	TradingSignal MUST have EXACTLY ONE of: position_pct, order_money, order_amount	halt
SL-06	filter_result column semantics: True=BUY, False=SELL, None/NaN=NO ACTION	halt
SL-07	Transformer MUST run BEFORE Accumulator in factor pipeline	halt
SL-08	MACD parameters locked: fast=12, slow=26, signal=9	halt

Full lock definitions: references/LOCKS.md

Top Anti-Patterns (15 total)

• AP-TIME-SERIES-ML-001: TimeSeries values array dimensionality mismatch
• AP-TIME-SERIES-ML-002: Non-floating-point dtype in TimeSeries values
• AP-TIME-SERIES-ML-003: Irregular or non-monotonic time index

All 15 anti-patterns: references/ANTI_PATTERNS.md

Evidence Quality Notice

[QUALITY NOTICE] This crystal was compiled from blueprint finance-bp-102. Evidence verify ratio = 43.8% and audit fail total = 26. Generated results may have uncaptured requirement gaps. Verify critical decisions against source files (LATEST.yaml / LATEST.jsonl).

Reference Files

File	Contents	When to Load
references/seed.yaml	V6+ 全量权威 (source-of-truth)	有行为/决策争议时必读
references/ANTI_PATTERNS.md	15 条跨项目反模式	开始实现前
references/WISDOM.md	跨项目精华借鉴	架构决策时
references/CONSTRAINTS.md	domain + fatal 约束	规则冲突时
references/USE_CASES.md	全量 KUC-* 业务场景	需要完整示例时
references/LOCKS.md	SL-* + preconditions + hints	生成回测/交易代码前
references/COMPONENTS.md	AST 组件地图（按 module 拆分）	查 API 时

---

Compiled by Doramagic crystal-compilation-v6.1 from finance-bp-102 blueprint at 2026-04-22T13:00:47.497902+00:00. See human_summary.md for non-technical overview.