Qlib Ai Quant

基于微软 qlib 的 AI 量化平台：覆盖预测模型、因子挖掘（Alpha158/TFT）、 组合优化、多频回测。支持 A 股 + 美股 + 港股多市场。

Security Scan

The skill claims to be a qlib-based quant platform but its runtime instructions mix qlib and ZVT workflows, reference undocumented environment variables and pip installs, and require filesystem writes — the pieces don't fully align and you should review before installing or running.

Purpose & Capability

Name/description claim a qlib-based AI quant platform, but the runtime files and seed.yaml include many ZVT-specific preconditions (python checks for zvt, ZVT_HOME directory, zvt.recorders commands) and references to multiple ecosystems. The required/env/installer metadata declares no env vars or installs, yet the instructions imply installing and using ZVT and other providers — this is disproportionate to a standalone 'qlib' helper and indicates incoherence between claimed purpose and required components.

Instruction Scope

SKILL.md/seed.yaml direct the agent to run runtime checks and commands (e.g., python3 -c 'import zvt...', pip install zvt if missing, create/check ZVT_HOME and touch files) and to reload seed.yaml before any behavioral decision. The skill's prose references reading/writing ~/.zvt and running recorders; it also directs the agent to consult many large reference files. These instructions access filesystem paths and environment variables (ZVT_HOME) that were not declared in requires.env and go beyond a simple 'write code for qlib' helper.

Install Mechanism

There is no declared install spec in registry metadata (instruction-only). However, seed.yaml/execution_protocol and SKILL.md imply installing packages (pip install zvt) and running host_adapter.install_recipes[]. The absence of an explicit install recipe in the registry while the instructions expect package installation is a mismatch and increases operational risk if the agent or user follows those steps automatically.

Credentials

Declared requirements list no environment variables or credentials, but the runtime instructions reference ZVT_HOME and require creating/writing to it. The skill also asks users to choose data providers (eastmoney, joinquant, qmt, etc.), some of which require API keys/accounts; those credentials are not declared. This mismatch means the skill expects access to filesystem locations and possibly external service credentials that were not declared up-front.

Persistence & Privilege

always:false (normal). The skill does instruct creating/checking a local data directory (~/.zvt) and suggests running pip install and recorders which will write local artifacts. It does not request to modify other skills or global agent config, but it does expect persistent local data directories and can cause environment changes if followed.

What to consider before installing

This skill's files claim a qlib-based platform but the runtime instructions heavily reference ZVT, expect Python package installs, and touch a local ZVT home directory — none of which are declared in the registry. Before installing or running: (1) Inspect SKILL.md and seed.yaml yourself and confirm you are comfortable with any pip installs; (2) run it inside an isolated virtual environment or sandbox and set ZVT_HOME to a dedicated directory to avoid contaminating your real ~/.zvt; (3) do NOT provide any API keys or credentials until you confirm which provider is actually used and why; (4) if you want to proceed, request the author/source (homepage is missing) or ask for an explicit install manifest and a minimal example showing only qlib usage — the current package is internally inconsistent and should be treated with caution.

Qlib AI 量化 (qlib-ai-quant)

用微软 qlib 做 AI 驱动的量化策略——预测模型、组合优化、Alpha158/TFT 特征工程，一套跑通。

Pipeline

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

Top Use Cases (38 total)

Multi-Frequency Data Resampling Instrument Processor (UC-101)

Resampling high-frequency 1-minute data to lower frequencies (e.g., daily) for downstream feature computation and model training Triggers: resample, frequency conversion, 1min

Specific Minute Selection Data Resampling (UC-102)

Resampling 1-minute data to daily frequency by extracting a specific minute point from each day for feature generation Triggers: resample, specific minute, 1min to day

Multi-Frequency Feature Handler (UC-103)

Loading and processing data with both daily frequency features and 15-minute frequency features for models that leverage multiple time scales Triggers: multi-frequency, 15min, day frequency

For all 38 use cases, see references/USE_CASES.md.

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 (25 total)

• AP-ZVT-183: 除权因子为 inf/NaN 时直接参与乘法导致复权静默失败
• AP-ZVT-179: 第三方数据接口超限后异常被吞噬，数据静默缺失
• AP-ZVT-183B: HFQ（后复权）与 QFQ（前复权）K 线表使用错误导致因子计算漂移

All 25 anti-patterns: references/ANTI_PATTERNS.md

Evidence Quality Notice

[QUALITY NOTICE] This crystal was compiled from blueprint finance-bp-087. Evidence verify ratio = 53.9% and audit fail total = 16. 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	25 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-087 blueprint at 2026-04-22T11:06:12.650493+00:00. See human_summary.md for non-technical overview.