Skill UI

A comprehensive guide for developers building AI demos and GPU compute tasks on Hugging Face Spaces using ZeroGPU hardware. It covers best practices for writing and reviewing code that utilizes the `@spaces.GPU` decorator, managing runtime duration and quota, and understanding the specific CUDA availability model and process isolation rules required for reliable ML deployment.

This skill is designed for complex, multi-step tasks requiring coordination among multiple AI agents. It implements a robust workflow engine that first generates a detailed plan (task graph, topology) and then orchestrates parallel execution with adversarial verification. Use this when the task spans many files, requires multiple specialized roles, or involves an overall 'workflow' concept, ensuring structured, reliable, and auditable results.

A comprehensive guide for training or fine-tuning advanced sentence-transformer models. This skill assists in selecting the correct architecture—including Bi-encoders (SentenceTransformer), CrossEncoders (for reranking), and SparseEncoders—for various Natural Language Processing (NLP) tasks. Use cases include semantic retrieval, similarity matching, document clustering, and pair-wise classification.

This skill provides expert capabilities for training and fine-tuning transformer language models using the TRL (Transformers Reinforcement Learning) library. It supports state-of-the-art post-training methods, including Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), Group Relative Policy Optimization (GRPO), and Reward Model training. Use this skill to align and customize foundation models for advanced tasks via CLI commands.

This meta-skill acts as a routing layer for complex UI development tasks. It guides the process of selecting the smallest, most specific, and useful skill context using the ui-skills CLI. Use it when starting UI work to analyze goals, identify the correct category, and pinpoint the precise skill set needed for implementation, ensuring development efficiency and accuracy.

This skill enforces a rigorous, multi-phase coding workflow (Discuss -> Map -> Decompose -> Execute -> Verify) to ensure code is structurally grounded, complete, and free from invented APIs or assumptions. It is designed for high-stakes, complex, or multi-file engineering tasks, acting as a senior, disciplined coding partner that plans meticulously before writing code.

A powerful plugin enabling Claude Code to delegate complex tasks to external AI CLIs (such as OpenAI Codex or Google Antigravity/Gemini) without exiting the conversation. It supports multi-model workflows, allowing users to seek a second opinion, cross-check deep analysis, or delegate side tasks (like research or code review). The core value is that Claude synthesizes the raw outputs, providing a critical review and reconciliation of findings across all consulted models.

This skill guides the setup of repeatable, self-running marketing workflows (loops). It transforms one-off tasks into 'always-on' systems that execute on defined cadences (daily, weekly, or by trigger). It is essential for defining the loop's purpose, matching the operational cadence to the underlying signal speed, and implementing necessary guardrails, self-checks, and human checkpoints to ensure sustained and efficient marketing operations.

This template provides a comprehensive three-screen UI showcase for a gamified habit tracking application, designed with an RPG (Role-Playing Game) aesthetic. It includes a main cover page, a 'Today's Quests' summary screen displaying XP and level progress, and a detailed quest view for managing subtasks and rewards. Perfect for showcasing personal growth, productivity, or educational concepts.

A highly visual Kanban board template designed for agile project and workflow management, similar to Trello. It features four distinct columns (To Do, In Progress, In Review, Done) to track the lifecycle of tasks. Users can filter by assignee, label, and search, and optionally group cards using swimlanes for enhanced clarity during operational sprints and team coordination.

A comprehensive Notion dashboard designed for real-time team operations and performance tracking. It provides a centralized overview by aggregating Key Performance Indicators (KPIs) in a grid format, displaying 7-day trend sparklines, showing live activity feeds, and summarizing tasks in a structured database table. Ideal for project managers and team leads who need a single source of truth for project status and team coordination.

The Agent Harness provides a robust, verifiable loop structure for driving complex goals across specialized domains. It converts a high-level objective into a verifiable task plan, executes tasks sequentially using domain-specific tools, and rigorously verifies every step with machine-run checks. The system includes mechanisms for retries, budget management, and mandatory human escalation, ensuring no goal is closed until all progress is verifiably complete or explicitly waived.