ml_engine

Triggered on runtime errors in the scan loop. Diagnoses the root cause, writes a targeted fix to src/, and reports what changed. The calling harness runs pytest before merging — if tests fail, the fix is reverted automatically.

Post-trade feedback loop system: shaped-reward logging, diagnostics, and self-heal. Use when investigating trade outcomes, checking system health, or understanding automated corrective actions.

Buddy MCP server: 15 tools for system observability, feedback loop control, OANDA state, and knowledge management. Use when querying system health, investigating trades, or triggering diagnostics/self-heal.

Spawned automatically on trade close by ClaudeReflectionHandler. Reads the just-closed trade, compares prediction to outcome, and writes structured learnings / rule drafts / weight proposals that Buddy picks up on the next scan cycle.

Triggered on runtime errors in the scan loop. Diagnoses the root cause, writes a targeted fix to src/, and reports what changed. The calling harness runs pytest before merging — if tests fail, the fix is reverted automatically.

Spawned automatically on trade close by ClaudeReflectionHandler. Reads the just-closed trade, compares prediction to outcome, and writes structured learnings / rule drafts / weight proposals that Buddy picks up on the next scan cycle.

Tune or audit ML Engine gate thresholds as a coordinated adaptive system. Use when confidence, momentum, risk, or agent-consensus thresholds are being changed; when one static gate bottlenecks an otherwise adaptive pipeline; or when the scanner alternates between over-trading and total suppression. Focus on coupled gate behavior, regime awareness, drawdown state, virtual trades, and safe one-change-at-a-time tuning.

Review changes to ML Engine's execution path, broker integration, and risk enforcement. Use when modifying `execution.py`, OANDA adapters, sizing, routing, or trade approval logic. Focus on preserving gate integrity, R:R minimums, correlation/risk controls, retry safety, state writes, and observability in a live trading environment.

Maintain the learning-to-rule pipeline in ML Engine. Use when learnings are malformed, promotions miscount patterns, archives bloat, mixed markdown formats conflict, or rules/learnings drift out of sync. Focus on compatibility across learnings producers, promotion safety, deduplication, and archive integrity.

Verify that a module, tracker, agent, or automation feature is truly wired into live production paths in ML Engine. Use when code exists but may be dead, when a phase claims completion, or when tests pass while runtime behavior suggests the feature is inert. Focus on config flags, production call sites, persistence writes, and runtime consumers.

Diagnose zero-trade stalls, chronic gate rejections, or sudden throughput collapse in the ML Engine trading loop. Use when the system stops producing trades, when tradeable pairs drop to zero, or when a user asks why the scanner is blocked. Focus on virtual trades, gate failures, threshold interactions, adaptive floors, and recent learnings/rules before changing code.

Buddy MCP server: 15 tools for system observability, feedback loop control, OANDA state, and knowledge management. Use when querying system health, investigating trades, or triggering diagnostics/self-heal.

Post-trade feedback loop system: shaped-reward logging, diagnostics, and self-heal. Use when investigating trade outcomes, checking system health, or understanding automated corrective actions.

Centralized configuration management with validation, versioning, and environment-specific overrides for the ML Engine FX trading system.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends the agent's capabilities with specialized knowledge, workflows, or tool integrations.