Skill UI

A robust framework for orchestrating complex, multi-step coding tasks using multiple AI agents. It functions by first planning the entire project into a mission dependency graph (DAG), then dispatching specialized agents to isolated Git worktrees for parallel development. The system automatically monitors progress, handles merges, and generates structured, comprehensive reports detailing changes, completed tasks, and next steps.

This tool routes tasks through a multi-tier system, specifically partitioning tasks into Agent Booster eligible (Tier 1) and standard LLM calls (Tier 2/3). It calculates and reports the estimated cost, leveraging the structural $0 cost of Booster bypasses for simple transforms, helping users optimize spend before running batches of similar tasks.

A deterministic 'Stop hook' designed to enforce session hygiene for AI agents. It checks machine-verifiable facts such as file modification timestamps of learning logs, disk space, and predefined rationalization patterns. This gate ensures that complex tasks lead to captured knowledge, acting as a mechanical safeguard complementing AI self-auditing.

A formal framework designed for evaluating AI-assisted code generation sessions. It implements Development Oriented Evals (EDD) principles, allowing users to define capability and regression criteria. The system supports multiple evaluation types—code-based, model-based, and human review—and tracks reliability using advanced metrics like pass@k, ensuring robust quality control for AI agents.

A comprehensive multi-agent feedback loop inspired by Generative Adversarial Networks (GANs), designed to build complex, production-ready applications. It separates generation from evaluation, using a 'ruthlessly strict' evaluator (QA) to drive continuous iteration. This process ensures high design quality, robust functionality, and originality, overcoming the limitations of single-agent self-critique. Ideal for full-stack feature development and complex product specification implementation.

This tool is designed to analyze source code by extracting core entities (classes, functions, modules, types) and mapping their complex relationships (imports, inheritance, calls, dependencies). It constructs a comprehensive Knowledge Graph (KG) in AgentDB. This is essential for detailed dependency analysis, understanding large codebase structures, and performing accurate impact assessments.

A comprehensive guide for optimizing latency-critical systems, such as market data feeds, real-time dashboards, and streaming agents. It details structured methods for mapping 'hot paths,' tracking key metrics (p95, p99 latency, throughput, freshness), and implementing architectural improvements like caching and batching to ensure high performance and data freshness under load.

This skill guides the design and rigorous review of goal-oriented agent feedback loops. It helps ensure that complex autonomous multi-step agents are robust, preventing failures such as infinite looping, goal-gaming, or running to completion with incorrect assumptions. It covers defining machine-decidable goals, setting boundaries, and selecting the correct loop type (servo, regulator).

A comprehensive, end-to-end methodology designed for software engineers and AI agents to safely and systematically integrate machine learning models into existing non-ML codebases. It guides users through critical phases, including problem framing, data readiness, architectural decoupling (via APIs/services), baseline model implementation, and final MLOps pipeline setup.

This comprehensive skill guides the user through the entire process of creating a deep, multi-layered AI persona (Soul.md). It helps define the identity, core rules, name, and visual prompts for agents designed for the OpenClaw platform. Users can initiate character creation via guided prompts or a random 'gacha' draw, ensuring a cohesive and compelling character profile.

This skill reviews prediction market, basket, oracle, and trading-agent workflows to ensure compliance, safety, data quality, and privacy. It must be run before any workflow handles user portfolio data, venue authentication, API keys, or trade planning to mitigate execution and regulatory risks across multiple domains.

A comprehensive security auditing tool designed for Claude Code environments. It scans the entire configuration directory for security vulnerabilities, misconfigurations, injection risks, hardcoded secrets, and overly permissive settings. Ideal for initial project setup, pre-commit checks, or routine security hygiene.