Claude Code skills for building reliable day trading bots: event-driven architecture, order execution integrity, position reconciliation, risk management, market data pipelines, backtesting, and fail-closed engineering practices. Encodes hard-won lessons from production trading bot failures.
Use when trading zero-days-to-expiration options, managing 0DTE gamma risk, implementing auto-exit rules near expiration, or handling SPY/SPX settlement differences.
Use when implementing asyncio tasks, background workers, or any concurrent code in trading systems, or when tasks fail silently or exceptions return None.
Use when implementing audit rules for trading systems, building forensic analysis for broker-vs-DB discrepancies, designing health score aggregation, or detecting silent failures across trading system components.
Use when building professional-grade backtesting frameworks, designing hypothesis-driven strategy tests, implementing walk-forward validation, or modeling realistic execution costs.
You MUST use this before any creative work - designing trading systems, adding strategies, modifying bot behavior, or building new features. Explores requirements and design before implementation.
Use when refactoring legacy trading code, when existing code lacks tests and behavior is unknown, or when you need to lock current behavior before making changes.
Use when parsing trading signals from chat messages, implementing multi-layer deduplication, routing signals by priority tier, or normalizing usernames and ticker mentions from Discord, Telegram, or web chatrooms.
Use when building a Chrome Extension to scrape trading signals from web chatrooms, designing browser-to-backend signal bridges, or implementing client-side message queuing and offline retry.
Use when setting up a new trading bot project with Claude Code, configuring CLAUDE.md for trading development, or establishing project-level rules and persistent memory for trading systems.
Use when implementing multi-model signal validation, ensemble scoring for trade decisions, or when adding confidence-based gates to prevent low-conviction trades.
Use when implementing database operations for trading systems, encountering transaction failures or stale data, or designing position and order persistence.
You MUST use this when encountering InFailedSqlTransaction errors, implementing nested database operations, using SAVEPOINT patterns, or designing transaction boundaries for concurrent trading systems.
Use when scaling a trading bot beyond single-process architecture, implementing event streaming with Kafka or NATS, decomposing into microservices, or building multi-bot coordination systems.
Use when running Playwright or Selenium scrapers in Docker, handling Cloudflare challenges, managing browser session persistence across restarts, or implementing health checks for headless Chrome.
You MUST use this when writing conditional checks on numeric trading data, handling missing market data values, or designing sentinel/default values for prices, VIX, premium, or equity.
Use when calculating market gaps, classifying gap types, implementing gap-based trading strategies, or when gap calculations produce incorrect results due to corporate actions.
Use when implementing IBKR TWS API connections, handling order ID management, bracket orders, reconnection logic, historical data pacing, orderStatus vs execDetails deduplication, or any IBKR API safety and edge-case handling.
Use when IBKR API orders are not submitting, sitting idle, or requiring manual TWS dialog confirmation — diagnoses TWS Precautionary Settings and IBC bypass config BEFORE debugging code.