Description

提供多市场金融强化学习环境，支持PPO/DQN等DRL算法回测、Markowitz组合优化与实时模拟交易，适配Alpaca等券商接口。。

README (SKILL.md)

FinRL 强化环境 (finrl-meta-envs)

Name: Finrl Meta Envs
Author: tangweigang-jpg

提供多市场金融强化学习环境，支持PPO/DQN等DRL算法回测、Markowitz组合优化与实时模拟交易，适配Alpaca等券商接口。

Pipeline

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

Top Use Cases (9 total)

Automated Paper Trading with PPO Agent (`UC-101`)

Execute simulated paper trades in real-time using a trained PPO reinforcement learning agent connected to Alpaca brokerage API, enabling risk-free str Triggers: paper trading, PPO agent, Alpaca

Alpaca Paper Trading Demo with PPO (`UC-104`)

Demonstrate live paper trading execution using a PPO neural network agent connected to Alpaca's paper trading API, enabling real-time trade simulation Triggers: paper trading, Alpaca demo, PPO

Markowitz Mean-Variance Portfolio Optimization (`UC-102`)

Optimize portfolio allocation across multiple assets using Markowitz mean-variance optimization to maximize risk-adjusted returns, balancing expected Triggers: portfolio optimization, Markowitz, mean-variance

For all 9 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-116. Evidence verify ratio = 23.2% and audit fail total = 8. 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-116 blueprint at 2026-04-22T13:00:56.369548+00:00. See human_summary.md for non-technical overview.

Usage Guidance

This skill looks like a legitimate FinRL/ZVT-style pipeline on paper, but its runtime instructions expect you to have and possibly install Python 3.12+, zvt and other libraries, create/check ~/.zvt, and connect to brokers like Alpaca — yet the registry lists no required env vars or install steps. Before installing or running it: 1) Ask the author to provide an explicit install spec and a list of required environment variables (Alpaca API_KEY/SECRET, ZVT_HOME, any DB credentials). 2) Inspect seed.yaml and the references folder yourself (they are included) for any commands that would run pip install or touch local files. 3) Run the skill only in an isolated environment (VM or Python venv) so automated pip installs or filesystem writes can't affect your primary workspace. 4) If you plan to connect to Alpaca or any broker, store keys in a vault or environment variables you control and confirm the skill will not transmit them elsewhere. 5) If you need higher assurance, request a concrete install script and minimal set of declared env vars, or decline until the author documents why no credentials are declared despite requiring broker/data access.

Capability Analysis

Type: OpenClaw Skill Name: finrl-meta-envs Version: 0.3.3 The skill bundle provides a legitimate and highly structured framework for financial reinforcement learning and backtesting using the ZVT and FinRL-Meta ecosystems. It includes extensive documentation on financial domain constraints, anti-patterns (e.g., AP-QLIB-1930), and semantic locks (e.g., SL-01) designed to prevent common modeling errors like look-ahead bias and survivorship bias. The execution protocol in 'references/seed.yaml' uses standard 'pip' installations for well-known libraries (numpy, pandas, torch, zvt), and the instructions are strictly aligned with the stated purpose of automated trading and factor research without any signs of malicious intent or data exfiltration.

Capability Tags

cryptocan-make-purchases

Capability Assessment

ℹ Purpose & Capability

The skill's name and description (FinRL-style multi-market RL, backtest, Markowitz, Alpaca paper trading) are coherent with the included use-cases and component references. However SKILL.md explicitly says it requires "Python 3.12+ with uv package manager" and references zvt/Alpaca/ZVT_HOME behaviors, while the registry metadata declared no required binaries, no env vars, and no install steps — a clear mismatch (the skill will realistically need Python and third‑party libs and likely brokerage API keys).

⚠ Instruction Scope

SKILL.md contains detailed runtime protocols and preconditions that tell the agent to run python3 -c checks (import zvt, call get_kdata), to create/check ~/.zvt (touch/unlink), and to re-read seed.yaml and references before execution. Those instructions direct the agent to read and write local filesystem paths and to run commands that may install packages (pip install) if preconditions fail. The instructions also reference Alpaca paper trading but do not declare the required Alpaca API keys or how to store them. The scope is broader than the registry claims and gives the agent discretion to run environment modifications.

ℹ Install Mechanism

No install spec or code files are present in the registry entry (instruction-only). That lowers the immediate risk of downloading arbitrary binaries. However SKILL.md and seed.yaml include an execution_protocol that tells the host to run install recipes and pip install commands if preconditions fail — these are not expressed as a formal install spec in the registry, so the agent (or a user following instructions) might perform installs later without explicit manifested approval.

⚠ Credentials

The registry lists no required environment variables or credentials, yet the SKILL.md and references clearly expect integration with zvt, a ZVT_HOME directory, and broker APIs (Alpaca) for paper trading. Alpaca (and likely other broker/data sources) require API keys/tokens; those are not declared in requires.env or primaryEnv. The skill also expects writable access to ~/.zvt. This under-declaration of secrets and environment requirements is disproportionate and could cause accidental credential use or ad-hoc prompts for secrets.

✓ Persistence & Privilege

always is false and there is no install spec that writes permanent binaries or modifies other skills. The seed.yaml/ SKILL.md ask the agent to re-read seed.yaml and run preconditions, but the skill does not request forced permanent inclusion or claim the ability to change other skills' configuration. Autonomous invocation is allowed (platform default) but does not by itself raise new flags here.

Version History

v0.3.3

v0.3.3: bilingual metadata injected. H1 shows FinRL 强化环境; tagline replaced with skill-specific Chinese hook; tags upgraded to Level 1-4.

v0.3.1

Remove install.sh — knowledge-only bundle. Host AI consumes directly from URL; no user-side installation needed. Fixes ClawHub suspicious flag.

v0.3.0

Doramagic crystal portfolio v0.3.0. Full 5-layer bp-009 standard. github.com/tangweigang-jpg/doramagic-skills

Metadata

Slug finrl-meta-envs

Version 0.3.3

License MIT-0

All-time Installs 0

Active Installs 0

Total Versions 3

Frequently Asked Questions

What is Finrl Meta Envs?

提供多市场金融强化学习环境，支持PPO/DQN等DRL算法回测、Markowitz组合优化与实时模拟交易，适配Alpaca等券商接口。。 It is an AI Agent Skill for Claude Code / OpenClaw, with 107 downloads so far.

How do I install Finrl Meta Envs?

Run "/install finrl-meta-envs" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Finrl Meta Envs free?

Yes, Finrl Meta Envs is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Finrl Meta Envs support?

Finrl Meta Envs is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Finrl Meta Envs?

It is built and maintained by Tang Weigang (@tangweigang-jpg); the current version is v0.3.3.

More Skills

Finrl Meta Envs