Skill UI

This tool ingests raw market data (OHLCV) for a specified symbol. It normalizes the data into relative OHLCV vectors, converts these vectors into high-dimensional embeddings, and then stores and indexes them using HNSW. This process prepares the data for fast, similarity-based pattern detection and advanced financial time series analysis.

This memory bridge ingests Claude Code's local auto-memory markdown files and structures them into AgentDB. It generates high-fidelity ONNX embeddings, handles deduplication using cosine similarity, and indexes the memories in a scalable namespace. This enables seamless, unified semantic search across all historical sessions and projects, significantly enhancing RAG capabilities.

This skill provides a structured framework for designing and implementing composable recommendation, ranking, and feed pipelines. It models the industry-standard six-stage pattern: Source, Hydrator, Filter, Scorer, Selector, and SideEffect. Use it for any system that selects "the top K items for a (user, context)", such as social feeds, ad ranking, RAG rerankers, or content recommendation engines, ensuring a robust and scalable architecture.

This skill is designed for complex, iterative decision-making processes requiring deep computation, such as stochastic optimization, high-dimensional search, or repeated model rollouts. It mandates the creation of a structured 'Ledger' to rigorously record every trial, prior winner, fresh information, and decision mark. It ensures that conclusions are based on durable evidence, managing risk through explicit promotion gates and preventing premature certainty.

A practical decision framework for determining whether to use regular expressions or Large Language Models (LLMs) when parsing structured text (e.g., forms, quizzes, invoices). It recommends starting with deterministic, low-cost regex for 95-98% of structured data, and only invoking expensive LLM calls for identifying and validating ambiguous edge cases, optimizing both cost and accuracy.

This is an advanced, structured workflow designed to conduct deep, multi-source investigations. It synthesizes current public evidence with user-provided context to generate highly structured reports. The workflow explicitly differentiates between verifiable facts, user-supplied evidence, logical inferences, and actionable recommendations, making it ideal for complex comparative analysis or continuous monitoring tasks.

This tool synthesizes multiple, scattered research findings retrieved from memory into a single, structured, and actionable report. It automatically grades evidence quality (High/Medium/Low), resolves conflicting claims, and provides concrete, recommendation-based insights, making complex data digestible for decision-making.

An advanced tool for network intelligence that calculates weighted graph scores to map the optimal path to a target. It identifies potential introducers, measures 'bridge value' across connections, and helps determine whether a connection requires a warm introduction or direct outreach on platforms like X and LinkedIn. Ideal for users needing the raw ranking methodology.

This skill executes comprehensive historical backtesting using the high-performance neural-trader engine. It supports features like walk-forward validation and captures key financial metrics (Sharpe, Drawdown). Critically, it signs the resulting artifact using Ed25519 cryptography, providing non-repudiable tamper evidence required for promoting research models to live trading systems.

This skill dispatches computationally intensive quantitative trading jobs (such as multi-year walk-forward analysis, large Monte-Carlo simulations, parameter sweeping, or deep model training) to a managed cloud agent runtime. It is designed for large-scale, production-grade backtesting that exceeds local machine capacity, ensuring robust execution and artifact management via Anthropic's cloud services.

This skill calculates model interpretability for algorithmic trading signals using PageRank. It constructs a feature-contribution graph from an LSTM or Transformer signal and executes a single-entry PageRank calculation. The result identifies the top-K features that most significantly drove the prediction, providing a "regulator-grade" audit trail essential for compliance with regulations like the EU AI Act.

Assess and manage investment risk using the neural-trader risk engine. This tool calculates critical metrics such as Value at Risk (VaR), Conditional VaR (CVaR), and Sharpe ratio for both single positions and entire portfolios. It also optimizes position sizing using methods like Kelly criterion and checks for market circuit breaker triggers (e.g., high correlation spikes or volatility regime shifts) to ensure robust trading decisions and minimize exposure.