Tradememory Protocol

TradeMemory Protocol is a memory layer for AI trading agents, published by Mnemox. It gives an agent persistent memory so it can record trades, recall similar past trades by outcome and explain why it acted, instead of forgetting everything when the context window ends. It exposes 20 tools over the Model Context Protocol and works with clients such as Claude Desktop. It is aimed at traders and developers running AI agents against markets, including equity traders and automated forex systems that sync from MT5. The workflow has three steps. Before trading, the agent recalls past trades in similar conditions with win rate and average profit. After trading, one call records the trade and updates five memory layers: episodic, semantic, procedural, affective and audit. Daily and weekly reviews look for behavioral drift, strategy decay and mistakes. An audit trail uses SHA 256 tamper detection, and the project lists compliance review as a design goal. The protocol is open source under the MIT license, runs on your own machine against a local SQLite database, and has no paid tier and no hosted service. The vendor offers free setup help. Separately, Mnemox sells strategy validation on its pricing page. Its Free plan costs $0 with 3 validations a month and basic reports, Pro costs $29 a month with unlimited validations, and Enterprise is custom priced for built to order trading infrastructure.