← 返回 Skills 市场
unrealbnb

Binance Coach

作者 UnrealBNB · GitHub ↗ · v2.1.2 · MIT-0
cross-platform ✓ 安全检测通过
643
总下载
1
收藏
2
当前安装
37
版本数
在 OpenClaw 中安装
/install binance-coach
功能描述
AI-powered crypto trading behavior coach for Binance users. Analyzes live portfolio health, detects emotional trading patterns (FOMO, panic selling, overtrad...
使用说明 (SKILL.md)

📊 BinanceCoach

Your AI-powered crypto trading behavior coach — connected to your Binance account.

BinanceCoach analyzes your live Binance portfolio, spots emotional trading patterns like FOMO and panic selling, and gives you smart DCA buy signals based on RSI and the Fear & Greed index — all via your OpenClaw assistant.


✨ What it does

Feature Description
💼 Portfolio Health Score 0–100 with grade, concentration warnings, stablecoin check
📐 Smart DCA Weekly buy amounts per coin, adjusted by RSI × Fear & Greed (25 combinations)
🧠 Behavior Analysis FOMO score, overtrading index, panic sell detector, streak tracker
📈 Market Context Live price, RSI, SMA50/200, trend direction per coin
😱 Fear & Greed Real-time index with buy/hold advice
🔔 Price Alerts Set price or RSI alerts, check when triggered
📚 Education 7 lessons: RSI, DCA, SMA200, Fear & Greed, concentration risk, panic selling
📅 Projections 12-month DCA accumulation projection per coin
📰 News & Listings Live Binance announcements, new coin listings, launchpools & HODLer airdrops
👁 Real-time Watcher Background daemon that polls Binance every 60s and Telegram-notifies instantly

🚀 Quick Start

Only one credential required when using with OpenClaw:

Binance API key + secret (read-only)

OpenClaw already has Claude built in and handles messaging — no Anthropic key or Telegram bot needed.

Just say: "analyze my portfolio" or "set up BinanceCoach" — your assistant handles the rest.

🔗 Optional: Hook into OpenClaw as default crypto handler

During setup, you'll be asked if you want BinanceCoach registered as your default handler for all crypto questions. If you say yes, a preference block is added to your OpenClaw USER.md:

  • Every session from that point on, your assistant uses BinanceCoach automatically for any crypto question
  • Covers: DCA, portfolio, market data, news, launchpools, listings, behavior analysis, alerts, projections, and more
  • You can remove it anytime by editing USER.md

This is opt-in — setup will always ask before modifying any config files.


🗣️ Example questions

  • "Analyze my portfolio"
  • "What's the Fear & Greed index?"
  • "Give me DCA advice for DOGE and ADA"
  • "Check my trading behavior"
  • "Set an alert if BTC drops below $60,000"
  • "Show me the market context for ETH"
  • "What's a 12-month DCA projection for BTC?"
  • "Explain dollar cost averaging"
  • "What are the latest Binance news / announcements?"
  • "Any new coin listings on Binance?"
  • "Are there any active launchpools or HODLer airdrops?"
  • "Watch for new Binance announcements and notify me instantly"
  • "Start the news watcher" / "Stop the watcher"

🔐 Security

  • Read-only Binance API — no trading, no withdrawals possible
  • Secrets stored locally only in ~/workspace/binance-coach/.env
  • .env is gitignored and never committed
  • Source code is fully auditable: github.com/UnrealBNB/BinanceCoachAI

See SECURITY.md for full audit guide.


⚙️ Optional: Standalone Telegram Bot

Want a dedicated Telegram bot (separate from OpenClaw)?

  1. Create a bot via @BotFather — send /newbot
  2. Get your Telegram user ID from @userinfobot
  3. Tell your OpenClaw: "Set up Telegram for BinanceCoach"

Your assistant will ask for the token and user ID, then start the bot.

The Telegram bot includes automatic news polling every 2 minutes — it notifies you instantly when new listings, launchpools, or announcements drop, without you having to ask.

👁 Real-time Announcement Watcher

Want instant Telegram notifications even outside the bot? Start the background watcher:

"Start watching for Binance announcements"

This runs a daemon that polls Binance every 60 seconds and pushes a Telegram message the moment something new appears — new listings, launchpools, HODLer airdrops, or news. It also flags announcements that affect coins you actually hold.


🌍 Languages

Supports English and Dutch (Nederlands). Switch anytime:

"Switch BinanceCoach to Dutch"


Built for the Binance OpenClaw AI Assistant Campaign 2026 · View on GitHub


\x3C!-- AGENT INSTRUCTIONS — read references/agent-guide.md for full dispatch table -->

安全使用建议
This skill appears internally consistent for a Binance portfolio coach, but review the following before installing: 1) Use a Binance API key with read-only permissions and optionally restrict it to your IP; never supply a trading-enabled key. 2) Installing runs 'pip install -r requirements.txt' (PyPI) — inspect requirements.txt if you need to audit third-party packages. 3) Secrets are written to ~/workspace/binance-coach/.env and are sourced by scripts (they become environment variables for the process); keep that directory secure and ensure .env remains gitignored. 4) Background watcher and scheduled crons will poll Binance and can send Telegram messages — only enable these if you want recurring notifications. 5) The skill can optionally append preferences to your USER.md — it will ask for consent; decline if you don’t want automatic handling of all crypto questions. If you want extra assurance, review the bundled source (src/) and requirements.txt before running setup.sh.
功能分析
Type: OpenClaw Skill Name: binance-coach Version: 2.1.2 BinanceCoach is a comprehensive crypto trading assistant that integrates Binance portfolio data with AI-driven coaching. It handles sensitive credentials (Binance API keys, Anthropic keys, and Telegram tokens) but follows security best practices by storing them locally in a .env file and providing clear instructions for creating read-only Binance keys. The skill includes a CLI wrapper (bc.sh) with basic shell injection safeguards, a Telegram bot with automated polling, and a setup wizard that requests explicit user consent before modifying the OpenClaw USER.md configuration. No evidence of data exfiltration, malicious execution, or unauthorized persistence was found; the code logic is consistent with its stated purpose of portfolio analysis and market monitoring.
能力评估
Purpose & Capability
The skill claims to analyze a Binance account and request BINANCE_API_KEY / BINANCE_API_SECRET — exactly the credentials needed. Required binaries (python3, pip3) and included Python sources implement portfolio, market, DCA, behavior, news, and Telegram features described in SKILL.md. Optional env vars (ANTHROPIC_API_KEY, TELEGRAM_*) are explained as optional for standalone modes.
Instruction Scope
Runtime instructions are focused on portfolio analysis, alerts, and an optional Telegram bot. They instruct writing Binance credentials to a local ~/.workspace/binance-coach/.env and creating optional crons / background watcher processes that poll Binance and (if configured) send Telegram messages. The instructions explicitly advise using read-only keys and not to echo secrets; they also tell agents to use the 'write' tool for secrets. This is appropriate, but the skill does create local files and can schedule recurring tasks that will run automatically if the user opts in.
Install Mechanism
Installer is bundled and copies source locally; the brew entry only ensures python3 exists. setup.sh runs 'pip install -r requirements.txt', which pulls packages from PyPI (python-binance, anthropic, requests, python-telegram-bot, etc.). That is normal for a Python tool but implies network fetches and execution of third‑party packages — moderate risk compared to instruction-only skills. No downloads from obscure URLs or shorteners detected.
Credentials
Only Binance API key + secret are required for core features; Anthropic and Telegram tokens are optional for standalone use. The amount and naming of environment variables are proportional to the functionality. The skill stores secrets locally in ~/.workspace/binance-coach/.env and sources that file at runtime (exporting to process env) — expected but worth noting because sourcing exports secrets into the running process environment.
Persistence & Privilege
always:false (normal). The skill offers to write a preference block to USER.md and to create OpenClaw crons and background watcher processes that poll Binance and push Telegram notifications — all are explicitly opt-in and require user consent during setup. These behaviors grant ongoing presence only when the user enables them; users should be aware that scheduled analyses and the watcher will run periodically and send portfolio data to Telegram if configured.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install binance-coach
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /binance-coach 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v2.1.2
binance-coach 2.1.2 - Updated scripts/daily-analysis.sh (details not shown). - No user-visible feature, config, or documentation changes outside code. - Documentation and feature set remain unchanged from previous version.
v2.1.1
binance-coach 2.1.1 - Updated daily-analysis.sh and daily_analysis.py scripts. - Minor maintenance and adjustments for daily analysis routines. - No changes to user-facing features or documentation.
v2.1.0
Version 2.1.0 introduces automated daily analysis and persistent behavioral tracking. - Added daily analysis scripts (`daily-analysis.sh`, `daily_analysis.py`) for automatic portfolio reviews. - Introduced `coach_db.py` for behavioral data storage and enhanced tracking of user patterns. - Improved Telegram bot handling and integrated new behavior analytics in `telegram_bot.py`. - Updated core modules to utilize new persistent storage and analysis features. - Enhanced market analysis, journaling, and alerts modules for seamless daily insights.
v2.0.0
**Major update with behavior, alerts, and codebase refactor** - Removed obsolete modules (`coach_db.py`, `history.py`, `_meta.json`) for a leaner codebase - Refactored core scripts and modules, improving maintainability and performance - Enhanced behavior analysis, price alert, DCA, and news functionality - Setup and execution scripts (`bc.sh`, `setup.sh`) updated - Documentation revised in `agent-guide.md` for improved guidance
v1.5.1
fix: security hardening — shell injection prevention, Telegram token log redaction, DB connection leak fixes, api_trade_count bug fix, StopIteration guard, real F&G in market history, non-interactive import-orders, bare except replaced with logger.debug
v1.5.0
feat: coach.db central source of truth — portfolio snapshots, DCA history, order import (opt-in), user action confirmations, today-vs-yesterday comparison
v1.4.3
fix: journal fallback for pnl + behavior when Binance API returns no trade history; extended window scan (365→730→1095d); behavior analysis now seeds from journal entries when API is empty
v1.4.2
fix: journal fallback for pnl + behavior when Binance API returns no trade history; extended window scan (365→730→1095d); behavior analysis now seeds from journal entries when API is empty
v1.4.1
fix v1.4.1: comprehensive code review — 27 bugs fixed: DB connection leaks, inflated P&L from ignored sells, falsy $0 P&L bug, HTML injection in Telegram, 90d→365d window, BNB commission nonsense, phantom BTC/ETH/BNB fallback, Python 3.9 compat, atomic targets write, untracked coins visible, /pnlexport sends file not path, /journaladd added to bot, all API failures logged not swallowed
v1.4.0
feat v1.4: Decision Journal (with OpenClaw memory sync), P&L Calculator (FIFO + CSV export), Portfolio Rebalancing Engine, Stablecoin Yield Optimizer — 4 high-impact features, 8 new commands, 8 new Telegram bot commands (29 total)
v1.3.5
fix: address security scanner flags — upfront setup disclosure, USER.md hook defaults to N, non-interactive mode skips USER.md entirely, update command shows what it fetches, removed inaccurate no-internet claim from README
v1.3.4
fix: USER.md hook is now opt-in — setup asks for explicit consent before modifying any config files
v1.3.3
binance-coach v1.3.3 - Updated metadata in _meta.json; no changes to core functionality or user-facing features. - All primary features, usage instructions, and documentation remain unchanged.
v1.3.2
BinanceCoach 1.3.2 - Default OpenClaw handler registration is now opt-in; setup always asks before modifying your config files. - Documentation updated to clarify you control integration with OpenClaw and can remove or avoid preference changes at any time. - No changes to core features or security practices.
v1.3.1
**Adds live Binance news, new coin listing, and announcement tracking features.** - Added live Binance news, listing, and announcement support, including HODLer airdrops and launchpools. - Introduced a real-time announcement watcher daemon with instant Telegram notification for major Binance updates. - Updated documentation and agent dispatch instructions to reflect news and watcher commands. - Improved OpenClaw integration: now auto-registers as default crypto skill after install. - Modernized/streamlined source and module structure; replaced and removed legacy/bot files.
v1.3.0
fix: 7 core improvements — behavior sync, panic sell detection, SMA200 in DCA multiplier, budget cap, portfolio-based default coins, working streak counters, per-asset projection scenarios
v1.2.1
fix: quote description field to prevent YAML parse error (colon in plain scalar caused parseFrontmatter to return empty object, silently discarding all metadata including requires.env)
v1.2.0
fix: match Notion-style multi-line metadata format; replace kind:download (unsupported) with kind:brew for python3; fix root cause of 'Required env vars: none' and 'No install spec' scanner flags
v1.1.9
fix: collapse metadata to single-line JSON per OpenClaw spec; add install spec; use requires.env + primaryEnv only
v1.1.8
fix: declare BINANCE_API_KEY/SECRET via requires.env + primaryEnv per OpenClaw skill spec — registry should now correctly show required credentials
元数据
Slug binance-coach
版本 2.1.2
许可证 MIT-0
累计安装 2
当前安装数 2
历史版本数 37
常见问题

Binance Coach 是什么?

AI-powered crypto trading behavior coach for Binance users. Analyzes live portfolio health, detects emotional trading patterns (FOMO, panic selling, overtrad... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 643 次。

如何安装 Binance Coach?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install binance-coach」即可一键安装,无需额外配置。

Binance Coach 是免费的吗?

是的,Binance Coach 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Binance Coach 支持哪些平台?

Binance Coach 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Binance Coach?

由 UnrealBNB(@unrealbnb)开发并维护,当前版本 v2.1.2。

💬 留言讨论