This release adds 5 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+5 more
Summary
AI summaryAll five memory types (Episodic, Semantic, Procedural, Affective, Prospective) are now fully implemented.
Full changelog
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
- Semantic — Bayesian updating for market beliefs
- Procedural — CUSUM drift detection for strategy rules
- Affective — confidence EWMA tracking
- Prospective — conditional plans with auto-induction
Platform-Agnostic Data Layer
DataSourceprotocol for pluggable market data- Binance adapter (live REST) + MT5 CSV adapter
- Context Builder for standardized OHLCV pipelines
Stats
- 15 MCP tools (was 10)
- 30+ REST endpoints
- 1,055 tests passing
- 62/62 roadmap tasks complete
Full changelog: CHANGELOG.md
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About mnemox-ai/tradememory-protocol
Structured 3-layer memory system (trades → patterns → strategy) for AI trading agents. Supports MT5, Binance, and Alpaca.
Related context
Beta — feedback welcome: [email protected]