agiprolabs-trading-skills

Claude Trading Skills

A comprehensive collection of 68 ready-to-use trading, DeFi, and quantitative finance Agent Skills. Works with Claude Code, Cursor, Codex, Gemini CLI, and 30+ other tools. Transform your AI agent into a trading analyst capable of executing complex multi-step workflows across market data analysis, on-chain research, backtesting, risk management, tax compliance, and more.

Crypto/DeFi-first. Extensible to all of quant finance.

🙌 Add your skills — pull requests welcome! This collection grows from the community's real trading and research experience. Built something useful? A new exchange or chain, a market type, a data source, a strategy, or a hard-won lesson — open a PR and share it. The bar is simple: a working SKILL.md in the Agent Skills format with honest, tested guidance (see CONTRIBUTING.md). Every contribution helps the whole community trade smarter.

🆕 Latest addition — prediction-market-live-ops: live-operations doctrine distilled from a real live-money Kalshi campaign — evidence-gated stage progression with pre-declared falsification lines, round-trip smoke tests, truthful cancels, empirical rate-limit probing, drawdown breakers, and why print-replay EV of passive fills is an upper bound, not an estimate. Ships with a companion open-source reference stack: kalshi-stack.

⚠️ Disclaimer: This is an analysis and research toolkit. Nothing produced by these skills constitutes financial advice. Always do your own research.

⭐ If you find this repository useful, please consider giving it a star! It helps others discover these tools and encourages continued development.

---

📋 Table of Contents

• Why Use This?
• What's Included
• Getting Started
• Quick Examples
• Prerequisites
• Available Skills
• Contributing
• Troubleshooting
• FAQ
• Citation
• License

---

🚀 Why Use This?

⚡ Accelerate Your Research

• Skip API boilerplate — Skills handle Birdeye, DexScreener, Jupiter, Helius, CoinGecko, DeFiLlama integration
• Production-ready code — Tested patterns for backtesting, risk management, and portfolio analytics
• Multi-step workflows — Chain data retrieval → analysis → visualization in a single prompt

🎯 Comprehensive Coverage

• 68 Skills across 17 categories covering the full trading workflow
• 7 Market Data APIs — Solana-native and cross-chain data sources
• 6 Solana Infrastructure tools for real-time streams, shreds, bundles, DEX aggregation, and transaction building
• 5 On-Chain Analysis tools for wallet profiling, whale tracking, sybil detection, and liquidity analysis
• Statistical + ML methods purpose-built for trading
• 7 Tax & Compliance tools for cost basis, wash sales, tax-loss harvesting, and reporting

🔧 Easy Integration

• One-click setup via Claude Code plugin or manual copy for any Agent Skills-compatible tool
• Automatic discovery — Your agent finds and uses relevant skills based on your request
• Extensible — Add equities, options, futures skills following the same pattern

---

📦 What's Included

Skill Categories

Category	Skills	Description
📊 Market Data & APIs	7	Birdeye, DexScreener, SolanaTracker, CoinGecko, Helius, Solana RPC, DeFiLlama
🔌 Solana Infrastructure	6	PumpFun mechanics, transaction building, Yellowstone gRPC, ShredStream, Jito bundles, Raptor DEX aggregator
🔗 On-Chain Analysis	5	Wallet profiling, holder analysis, whale tracking, liquidity, sybil detection
📈 Technical Analysis	3	pandas-ta, TA-Lib, crypto-native custom indicators
🔄 Backtesting & Strategy	4	vectorbt, Backtrader, strategy framework, walk-forward validation
⚖️ Portfolio & Risk	4	Portfolio analytics, position sizing, risk management, Kelly criterion
🏦 DeFi Specific	6	LP math, impermanent loss, yield analysis, MEV, tokenomics, DEX pool analysis
📐 Statistical Methods	5	Regime detection, volatility modeling, cointegration, mean reversion, correlation
🤖 ML for Trading	4	Signal classification, feature engineering, RL execution, sentiment
⚡ Execution & Trading	4	DEX execution, slippage modeling, copy-trading, exit strategies
📉 Data & Visualization	3	Trading charts, OHLCV processing, trade journaling
🔬 Market Microstructure	2	DEX orderflow analysis, traditional LOB theory, market making
🔮 Quant Finance	2	Options pricing, fixed income
🗳️ Prediction Markets	6	Kalshi & Polymarket exchange APIs, weather + crypto/index range markets, the strategy/sizing/backtesting layer, and live-operations doctrine from a real campaign
🧾 Tax, Accounting & Compliance	7	Cost basis, wash sales, tax-loss harvesting, exports, bookkeeping, reporting

Each skill includes:

• ✅ Comprehensive documentation (SKILL.md)
• ✅ Code examples with working snippets
• ✅ Use cases and integration guidance
• ✅ Reference materials in references/

---

🎯 Getting Started

This repo follows the open Agent Skills standard and works with any compatible agent.

Option A: Claude Code Plugin (Recommended)

The fastest way to install. Requires Claude Code v1.0.33+.

Step 1: Add the marketplace

/plugin marketplace add agiprolabs/claude-trading-skills

Step 2: Install the plugin

/plugin install trading-skills@agiprolabs-claude-trading-skills

That's it! All 68 skills are now available. Claude will automatically discover and use them when relevant to your trading tasks. Skills are namespaced as /trading-skills:skill-name.

Managing the plugin:

/plugin                                                    # Browse installed plugins
/plugin disable trading-skills@agiprolabs-claude-trading-skills   # Disable
/plugin enable trading-skills@agiprolabs-claude-trading-skills    # Re-enable
/plugin uninstall trading-skills@agiprolabs-claude-trading-skills # Remove
/reload-plugins                                            # Apply changes without restart

Option B: Manual Copy (Any Agent Skills-Compatible Tool)

Works with Claude Code, Cursor, Codex, Gemini CLI, and 30+ other tools.

Step 1: Clone the repository

git clone https://github.com/agiprolabs/claude-trading-skills.git

Step 2: Copy skills to your agent's skills directory

Scope	Claude Code	Cursor	Codex	Gemini CLI
Global (all projects)	~/.claude/skills/	~/.cursor/skills/	~/.codex/skills/	~/.gemini/skills/
Project (one project)	.claude/skills/	.cursor/skills/	.codex/skills/	.gemini/skills/

# Global install — choose your tool:
cp -r claude-trading-skills/skills/* ~/.claude/skills/   # Claude Code
cp -r claude-trading-skills/skills/* ~/.cursor/skills/   # Cursor
cp -r claude-trading-skills/skills/* ~/.codex/skills/    # Codex
cp -r claude-trading-skills/skills/* ~/.gemini/skills/   # Gemini CLI

# Or project-level install (Claude Code example):
mkdir -p .claude/skills
cp -r /path/to/claude-trading-skills/skills/* .claude/skills/

Your agent will automatically discover the skills and use them when relevant to your trading tasks.

Note: Agent Skills are supported by 30+ tools including Claude Code, Cursor, Codex, Gemini CLI, VS Code Copilot, Roo Code, Goose, OpenHands, and more. See agentskills.io for the full list.

---

💡 Quick Examples

🔍 Token Deep Dive

Use available skills. Analyze this Solana token: [MINT_ADDRESS]. Get current price and volume 
from Birdeye, check holder distribution and top 10 concentration, analyze liquidity depth across 
DEX pools, compute RSI/MACD/BBands on 1h candles, estimate slippage for a 0.5 SOL entry, and 
give me a risk assessment with suggested position size assuming 2% portfolio risk.

Skills Used: birdeye-api, token-holder-analysis, liquidity-analysis, dex-pool-analysis, pandas-ta, slippage-modeling, risk-management, position-sizing

📊 Strategy Backtest

Use available skills. Backtest a simple RSI mean-reversion strategy on SOL/USDC 1h candles for 
the past 6 months. Entry when RSI(14) < 30, exit when RSI > 70. Use 1% risk per trade with 
ATR(14)*2 stop loss. Show me equity curve, drawdown chart, and key metrics (Sharpe, max DD, 
win rate, profit factor).

Skills Used: birdeye-api, ohlcv-processing, pandas-ta, vectorbt, portfolio-analytics, trading-visualization

🏦 DeFi Yield Comparison

Use available skills. Compare yield opportunities for SOL across Raydium, Orca, and Meteora 
LP pools. Calculate impermanent loss at ±25% and ±50% price moves. Show me net APY after IL 
for each pool, and rank them by risk-adjusted yield. Include MEV risk assessment.

Skills Used: defillama-api, dex-pool-analysis, lp-math, impermanent-loss, yield-analysis, mev-analysis, trading-visualization

🐋 Whale Monitoring

Use available skills. Analyze the top 20 holders of [TOKEN_MINT]. Identify which wallets have 
been accumulating in the past 7 days. Check if any are known profitable traders. Show me the 
holder distribution with a Gini coefficient and flag any concentration risks.

Skills Used: helius-api, token-holder-analysis, whale-tracking, solana-onchain, trading-visualization

🧾 Tax-Aware Position Management

Use available skills. I'm tracking cost basis for my SOL trades using proportional (average)
cost basis for my accumulate/house-money strategy. I bought 10 SOL at $150, then 5 SOL at $180.
Now I want to sell 3 SOL at $200. Calculate my cost basis, realized gain, and remaining position.
Then check if any of my recent sells trigger wash sale rules, and show me what my Form 8949
entries would look like. Export the results in Koinly-compatible format.

Skills Used: cost-basis-engine, tax-liability-tracking, wash-sale-detection, crypto-tax-export, regulatory-reporting

🤖 ML Signal Pipeline

Use available skills. Build a feature set for [TOKEN] using 1h OHLCV data: RSI, MACD, 
Bollinger Band width, volume ratio, and holder count momentum. Train an XGBoost classifier 
to predict >2% returns in the next 4 hours. Use walk-forward validation with 30-day train, 
7-day test windows. Show feature importance and classification metrics.

Skills Used: birdeye-api, ohlcv-processing, pandas-ta, custom-indicators, feature-engineering, signal-classification, trading-visualization

---

⚙️ Prerequisites

• Python: 3.9+ (3.12+ recommended)
• uv: Python package manager (required for installing skill dependencies)
• Client: Any agent that supports the Agent Skills standard (Claude Code, Cursor, Codex, Gemini CLI, etc.)
• System: macOS, Linux, or Windows with WSL2
• API Keys (as needed):
  • Birdeye (free tier at birdeye.so)