# tradememory-protocol

> Decision audit trail + persistent memory for AI trading agents. Outcome-weighted recall, SHA-256 tamper detection, 17 MCP tools.

- **URL**: https://www.freshcrate.ai/projects/tradememory-protocol
- **Author**: mnemox-ai
- **Category**: MCP Servers
- **Latest version**: `v0.5.1` (2026-03-27)
- **License**: MIT
- **Source**: https://github.com/mnemox-ai/tradememory-protocol
- **Homepage**: https://mnemox.ai/tradememory/
- **Language**: Python
- **GitHub**: 647 stars, 94 forks
- **Registry**: github
- **Tags**: `ai-agents`, `claude`, `crypto`, `evolution-engine`, `forex`, `mcp`, `mcp-server`, `memory`, `python`

## Description

Decision audit trail + persistent memory for AI trading agents. Outcome-weighted recall, SHA-256 tamper detection, 17 MCP tools.

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v0.5.1` | 2026-03-27 | Medium | ## What's New  ### Decision Audit Trail (TDR) - 4 audit REST endpoints: `/audit/decision-record/{id}`, `/audit/export`, `/audit/export-jsonl`, `/audit/verify/{id}` - Trading Decision Record schema inspired by MiFID II Article 17 + EU AI Act Article 14 - SHA-256 tamper detection on every record  ### DecisionLogReader - Reads EA JSONL decision logs (conditions, filters, indicators, execution, regime, risk) - Priority: JSONL > CSV event_log > fallback - Handles concurrent write corruption (merged J |
| `v0.5.0` | 2026-03-16 | Low | ## What's New  ### Evolution Engine - Autonomous strategy discovery powered by LLM-generated hypotheses - Vectorized backtester for fast signal evaluation - In-sample / out-of-sample validation to prevent overfitting - Multi-generation evolution: discover → hypothesize → backtest → select → evolve - Result: BTC/USDT 1H over 22 months — Sharpe 3.84, 477 trades, 91% positive months  ### OWM Complete All 5 memory types fully implemented: - **Episodic** — power-law decay for trade recall |
| `v0.4.0` | 2026-03-05 | Low | ## Outcome-Weighted Memory (OWM)  A cognitive science-based recall system for AI trading agents. Novel application of established cognitive science (ACT-R, Kelly Criterion, Bayesian updating, Tulving's memory taxonomy) to AI trading agents.  ### New Features  - **5 Memory Types**: Episodic, Semantic, Procedural, Affective, Prospective - **Core Formula**: `Score(m,C) = Q(m) × Sim(m,C) × Rec(m) × Conf(m) × Aff(m)` - **Kelly-from-Memory**: Context-weighted position sizing derived from recalled trad |
| `v0.3.1` | 2026-03-02 | Low | ## [0.3.1] - 2026-03-03  ### Added - `scripts/generate_screenshots.py` — generates demo output for documentation - `ROADMAP.md` — 5-phase development roadmap - OpenClaw Skill (`.skills/tradememory/SKILL.md`) with env var declarations and security section - Hosted API server (`hosted/server.py`) with account isolation and API key auth - Marketing materials (`marketing/`) for Forex Factory, Reddit, MQL5  ### Changed - Repository reorganized: scripts, docs, and deploy configs moved to pro |
| `v0.3.0` | 2026-03-01 | Low | ## L3 Strategy Adjustments — Rule-based strategy tuning from L2 patterns  ### Added - `strategy_adjustments` table in SQLite with proposed/approved/applied/rejected lifecycle - 5 deterministic rules: strategy_disable, strategy_prefer, session_reduce, session_increase, direction_restrict - `generate_l3_adjustments()` in ReflectionEngine — reads L2 patterns, outputs proposed adjustments - 3 CRUD methods in Database: `insert_adjustment`, `query_adjustments`, `update_adjustment_status` - 3 REST API |
| `v0.2.0` | 2026-02-24 | Low | ## tradememory-protocol v0.2.0  The first MCP memory system built for AI trading agents. Not a dashboard. Not a journal. A protocol layer that gives your trading AI the ability to remember and learn.  ### New: MCP Server (FastMCP)  4 tools available via MCP protocol:  \| Tool \| Description \| \|------\|-------------\| \| `store_trade_memory` \| Store a trade decision with full context \| \| `recall_similar_trades` \| Find past trades with similar market context \| \| `get_strategy_performance` \| Aggregate s |
| `v0.1.0` | 2026-02-23 | Low | ## TradeMemory Protocol v0.1.0  The first public release of TradeMemory Protocol — a structured memory layer for AI trading agents.  ### Features - 3-layer memory architecture (L1: Raw Trades → L2: Patterns → L3: Strategy) - 7 MCP tools for recording, reflecting, and evolving trading decisions - ReflectionEngine powered by Claude API - Platform adapters: MT5, Binance, Alpaca - 36 tests passing, zero deprecation warnings  ### Quick Start See [QUICK_START.md](docs/QUICK_START.md) for a |

## Citation

- HTML: https://www.freshcrate.ai/projects/tradememory-protocol
- Markdown: https://www.freshcrate.ai/projects/tradememory-protocol.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/tradememory-protocol/deps

_Generated by freshcrate.ai. Indexes github releases for AI-agent ecosystem packages._