Tensortrade Rl Env

提供多市场回测与强化学习交易环境构建能力，支持多交易所钱包组合管理、Plotly交互式交易可视化及RL智能体训练评估。

This skill looks like domain-appropriate documentation for building RL trading/backtest pipelines, but there are a few red flags you should address before running it: - Source trust: the skill's source/homepage is unknown and the license is proprietary — only proceed if you trust the publisher. - Declared vs. actual requirements: SKILL.md requires Python 3.12+ and the 'uv' package manager and references the ZVT runtime (and the ZVT_HOME env var), but the registry metadata lists no required binaries, env vars, or config paths. Treat that as a mismatch—set up a clean environment yourself (see below) rather than letting an agent auto-install. - File writes & installs: the skill's preconditions will attempt to pip-install zvt and create/write ~/.zvt; do not run these steps on a sensitive machine. Use a dedicated virtualenv/container (Python 3.12) or sandbox. - Credentials: examples mention multi-exchange wallets (Bitfinex/Bitstamp); do NOT supply real exchange API keys unless you understand exactly how the skill uses them and you trust the code — prefer simulated wallets first. The skill does not declare any required KEY/TOKEN vars, so any prompt for secrets should be treated suspiciously. - Review important files: inspect references/LOCKS.md, references/seed.yaml and references/ANTI_PATTERNS.md yourself to understand fatal semantic checks (e.g., next-bar execution, T+1 rules) before running automated backtests. Recommended safe steps: run in an isolated VM/container; create a Python 3.12 virtualenv; set ZVT_HOME to a disposable directory you control; inspect the seed.yaml and SKILL.md contents locally; run only read-only prechecks first (import checks) and refuse automatic pip installs unless you audit the packages and trust PyPI sources. If you must run production-style backtests, prepare simulated data and avoid providing any live-exchange credentials until the code provenance is verified.

Type: OpenClaw Skill Name: tensortrade-rl-env Version: 0.3.3 The bundle is a highly structured framework for developing reinforcement learning trading strategies using the ZVT and TensorTrade libraries. It contains extensive domain-specific safety mechanisms, such as 'Semantic Locks' (e.g., SL-01, SL-02) and 'Fatal Constraints' (e.g., finance-C-001), designed to prevent common quantitative trading errors like lookahead bias, survivorship bias, and financial rounding errors. The installation recipes and preconditions are standard for the OpenClaw environment, and the instructions focus entirely on enforcing financial logic and data integrity without any evidence of malicious intent, data exfiltration, or unauthorized execution.