Skill UI

This workflow guides developers through Test-Driven Development (TDD) principles. It mandates writing tests (unit, integration, E2E) before writing implementation code, ensuring minimum 80% test coverage. The structured process (Red -> Green -> Refactor) guarantees robust code quality, effective bug fixing, and safe feature addition, while integrating Git checkpoint commits for full traceability.

This skill provides a structured framework for managing complex, multi-agent development tasks. It treats agents as a collaborative team, utilizing structured work items, defined ownership, Agent Kanban boards, and explicit merge gates. This ensures that large-scale, multi-tool, or cross-branch development efforts are predictable, highly visible, and result in reliably mergeable, production-ready artifacts.

This skill intercepts the response flow before generation, allowing the user to explicitly control the desired depth, length, and token usage of the AI's answer. It estimates the complexity of the input prompt and presents multiple options (e.g., 25%, 50%, 75%, 100%), ensuring the output matches the user's need, whether they require a brief summary or a comprehensive, detailed analysis.

Performs comprehensive historical backtesting for trading strategies using the high-performance neural-trader engine (Rust/NAPI). This skill supports walk-forward validation and captures key performance metrics. Critically, it Ed25519-signs the final artifact to provide cryptographic tamper evidence, ensuring data integrity for promotion from paper to live trading environments.

Dispatches computationally intensive quantitative financial jobs (such as multi-year walk-forward backtesting, large Monte-Carlo simulations, parameter sweeps, and deep model training using neural-trader) to a managed cloud agent. This is necessary for workloads that exceed local computational limits, providing robust, scalable infrastructure for professional quantitative research.

This skill provides regulator-grade feature attribution for complex trading signals (LSTM/Transformer). By constructing a feature-contribution graph and running single-entry PageRank, it identifies the top-K features that most drove a prediction. This is crucial for compliance with regulations like the EU AI Act and SEC Reg-AI, providing a robust, auditable audit trail for algorithmic trading systems.

This tool automates the entire process of optimizing investment portfolio allocations using a sophisticated mean-variance engine. Users can set specific risk targets, assess detailed risk metrics (variance, correlation), predict expected returns using neural networks, and generate actionable, step-by-step rebalancing plans. It is designed for quantitative traders managing complex asset baskets and historical pattern analysis.

This skill solves the mean-variance portfolio optimization problem (Σ · x = μ) using the highly efficient Conjugate Gradient (CG) method. It significantly outperforms legacy Neumann series methods, providing state-of-the-art speedup (40-60x) and accuracy for calculating optimal investment weights. It is designed for quantitative finance and risk management applications where computational performance is critical.

This tool detects the current market regime (e.g., bull, bear, ranging, volatile) by leveraging advanced neural network models and technical indicators. It provides a classification, confidence score, and a recommended trading strategy, making it essential for informed decision-making, strategic planning, and enhancing the context before running complex backtests or signals.

A comprehensive tool for assessing both portfolio and single-position risk using the neural-trader risk engine. It calculates critical metrics such as VaR, CVaR, and Sharpe Ratio. Furthermore, it provides optimal position sizing recommendations (e.g., Kelly criterion) and checks for market stress signals via circuit breaker status (e.g., loss limits, correlation spikes). Ideal for quantitative analysis and risk management in trading.

This skill generates sophisticated trading signals by employing the neural-trader anomaly detection engine. It utilizes Z-score scoring to classify market movements into distinct patterns, such as spikes (breakout), drifts (trend forming), and oscillations (range-bound). The process integrates neural prediction and historical pattern searching to provide highly ranked, actionable signals, making it invaluable for algorithmic and quantitative trading strategies.

This skill provides a comprehensive workflow for training advanced neural network models, including LSTM, Transformer, and N-BEATS, using historical market data. It facilitates model training, prediction generation with confidence intervals, performance comparison across different model types, and the structured storage of results for further analysis and SONA training.