Skill UI

A comprehensive, adaptive framework for developing and implementing features, adhering to a strict, multi-phase workflow: Specify, Design, Tasks, and Execute. This skill ensures feature quality through granular task breakdown, atomic Git commits, requirement traceability, and mandatory, independent verification. It automatically adjusts its depth based on feature complexity (Small to Complex), providing a robust mechanism for managing technical debt and capturing project lessons.

Workthreads analyzes a team's coding history across multiple repositories to generate a structured weekly status report. It clusters raw logs into 'work threads' and automatically tags their lifecycle status (new, open loops, closed, reverted). This is essential for leads needing a high-level, actionable summary of project progress, outstanding tasks, and completed features without reading thousands of lines of raw activity.

This advanced protocol forces the AI to conduct a thorough context audit before attempting any complex task. It mandates reading every relevant file, identifying architectural gaps, and providing a detailed orientation statement. It is essential for multi-file modifications, system design, and debugging to prevent acting on assumptions and maintain high code accuracy. The AI cannot proceed until all unknown variables are resolved.

Tamarind Bio is a cloud platform providing managed GPU access to hundreds of open-source computational biology tools. It facilitates complex tasks such as protein structure prediction (AlphaFold), de novo protein design, antibody engineering, protein-ligand docking, and molecular dynamics simulations. Users submit jobs via a unified API and receive predicted structures and biophysical scores without needing local high-performance computing resources.

This skill provides utilities to manage and drain the deliberation queue (brain_frames) of the ACC system. It allows users to list waiting frames, retrieve necessary context, perform deliberation on open tasks, and submit continuations using `acc_act(runtime="continue")`. It is essential for checkpointed work saved by headless runs, ensuring continuous task resolution and calibration against predicted outcomes in complex AI workflows.

This skill performs a comprehensive usability audit and heuristic evaluation of interactive user interfaces (UIs). It goes beyond mere visual aesthetics, assessing whether a system is intuitive, easy to learn, and functional for a first-time user. Use it when auditing complex tools, editors, or apps to ensure the user can complete primary tasks without needing extensive instructions or guidance.

Use this comprehensive skill for interacting with the Gemini API. It supports a wide range of advanced AI tasks, including text generation, multi-turn chat, multimodal understanding, image/video creation, function calling, and structured output. Provides modern SDKs for robust application development in Python and JavaScript/TypeScript.

This professional tool helps developers and researchers identify the optimal AI model for a specific task, such as coding, reasoning, or image recognition. It queries official HuggingFace leaderboards, filters models based on hardware constraints (e.g., available VRAM or RAM), and provides a detailed comparison table including benchmark scores, parameter size, and licensing information.

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.