TradeMemory Plugin

Persistent memory + autonomous strategy evolution for AI traders. 200+ trading MCP servers execute. None remember. TradeMemory does.

Installation

From GitHub (recommended)

git clone https://github.com/mnemox-ai/tradememory-plugin.git
claude --plugin-dir ./tradememory-plugin

Manual

Copy the plugin directory into your project or pass it directly:

claude --plugin-dir /path/to/tradememory-plugin

MCP Server (standalone, no plugin needed)

pip install tradememory-protocol
claude mcp add tradememory -- uvx tradememory-protocol

Commands

Command	Description
/record-trade [details]	Record a completed trade into all 5 OWM memory layers
/recall [context]	Recall similar past trades, ranked by outcome-weighted score
/performance [strategy]	Generate strategy performance report with behavioral analysis
/evolve [symbol] [tf] [gens]	Discover new trading strategies from raw OHLCV data
/daily-review [date]	AI-powered daily reflection on trades and behavioral patterns

Skills

Trading Memory

Skill	Description
trading-memory	OWM architecture, 5 memory types, recall scoring, behavioral baselines
evolution-engine	LLM-powered strategy discovery, vectorized backtesting, OOS validation
risk-management	Affective state monitoring, tilt detection, position sizing, behavioral guardrails

MCP Tools (15 total)

Core Memory (4)

• store_trade_memory — Store a trade with context
• recall_similar_trades — Find past trades matching current context
• get_strategy_performance — Aggregate stats per strategy
• get_trade_reflection — Deep-dive into a trade's reasoning

OWM Cognitive Memory (6)

• remember_trade — Store across all 5 OWM memory layers
• recall_memories — Outcome-weighted recall
• get_behavioral_analysis — Disposition ratio, hold times, Kelly criterion
• get_agent_state — Confidence, drawdown, streaks, risk appetite
• create_trading_plan — Prospective trading plans
• check_active_plans — Evaluate plans against current conditions

Evolution Engine (5)

• evolution_fetch_market_data — Fetch OHLCV from Binance
• evolution_discover_patterns — LLM-powered pattern discovery
• evolution_run_backtest — Vectorized backtesting
• evolution_evolve_strategy — Full evolution loop
• evolution_get_log — Evolution history and graveyard

Example Workflows

Record and Learn

/record-trade XAUUSD long 5180 5210 +$150

# Stores trade, updates all memory layers, shows similar past trades

Pre-Trade Check

/recall London session breakout, high volatility, XAUUSD trending up

# Returns past trades in similar conditions, ranked by P&L outcome

Strategy Evolution

/evolve BTCUSDT 1h 3

# Discovers patterns → backtests → selects → mutates × 3 generations
# Validates out-of-sample → graduates survivors

End of Day

/daily-review today

# Analyzes today's trades, checks behavioral drift, updates affective state

Requirements

• Python 3.10+
• pip install tradememory-protocol
• Optional: ANTHROPIC_API_KEY for LLM reflections and Evolution Engine

Links

• Plugin repo
• Core protocol
• PyPI
• Tutorial