skills-based-app

This skill should be used when users need help with Git, GitHub workflows, pull requests, branching strategies, repository contributions, and open source collaboration.

FastAPI Python backend for LLM Arena game server. Use when working on API endpoints, LLM provider integration via LiteLLM, game move processing, prompt engineering for AI players, error handling, or backend testing.

React 19 + Vite frontend for LLM Arena game visualization. Use when working on UI components, game board rendering, state management with hooks, API integration, Tailwind CSS v4 styling, Framer Motion animations, or frontend build configuration.

Full-stack LLM Arena application for AI vs AI game battles. Use when working on project-wide configurations, deployment, architecture decisions, or coordinating between frontend and backend. Covers Tic-Tac-Toe and Reversi games with OpenAI, Anthropic, and Google Gemini support.

Audits mobile app codebases for pre-release readiness across iOS (Swift/SwiftUI/UIKit), Android (Kotlin/Jetpack Compose), Flutter (Dart), and React Native. Produces a ship / ship-with-caveats / block verdict backed by findings in six layers — privacy & permissions, accessibility, localization, crash-prone code, store metadata, and release-build hygiene. Use when given a mobile app repo (or parts of one) and asked to review, audit, pre-flight, check if ready to ship, prepare for App Store or Play Store submission, run a release checklist, do an accessibility or privacy compliance review, or diagnose a likely store rejection. Also use when user mentions TestFlight, beta submission, privacy manifest, PrivacyInfo.xcprivacy, Data Safety form, ATT prompt, content rating, age rating, export compliance, Info.plist audit, AndroidManifest audit, or "why will this get rejected".

Assistant for designing, building, and optimizing Power BI semantic models. Helps with data architecture decisions, DAX formula development, relationship configuration, performance tuning, and implementation of analytics best practices. Triggers on questions about model design patterns, metric calculations, schema optimization, data integrity, security frameworks, and documentation standards.

Use this skill when the user asks to load, fine-tune, or run inference with pre-trained transformer models using the Hugging Face transformers library. Triggers include requests like "classify this text with BERT", "fine-tune GPT-2 on my dataset", "build a text summarization pipeline", "tokenize text for transformers", or "run question answering with a pre-trained model". Also use when the user asks about AutoTokenizer, AutoModel, Trainer API, or pipeline(). Do NOT use for CNNs, RNNs, NLTK, spaCy, or tasks that don't involve transformer architecture or the Hugging Face library.

Use this skill when the user asks to clean, preprocess, tokenize, stem, lemmatize, or vectorize raw text data using Python's NLTK library. Triggers include requests like "clean this text dataset", "tokenize my corpus", "build an NLP preprocessing pipeline", "convert text to TF-IDF or Bag of Words", "compare stemming vs lemmatization", or "evaluate my NLP model". Also use when the user asks about stopword removal, punctuation stripping, frequency analysis, or building scikit-learn compatible text pipelines with NLTK. Do NOT use for deep learning NLP (transformers/BERT/spaCy) or for tasks that don't involve text preprocessing.

Statistical visualization. Scatter, box, violin, heatmaps, pair plots, regression, correlation matrices, KDE, faceted plots, for exploratory analysis and publication figures.

Pandas EDA skill - Covers data loading, cleaning, transformation, aggregation, and visualization workflows used in real-world data analysis.

Analyzes customer support chats, tickets, and messages to compute satisfaction metrics, detect frustration, score response quality, predict churn risk, and suggest the best next action. Use when given support conversations, customer messages, WhatsApp threads, or tickets and asked to evaluate, score, improve, or act on them. Also use for CSAT, NPS, CES analysis, escalation decisions, or agent performance reviews.

Learn how to compress machine learning models using distillation, pruning, and quantization for real-world deployment.

Use this skill when the user wants to create, write, improve, or review a SKILL.md file for an AI agent. Triggers include any mention of "skill file", "skill.md", "agent skill", "writing a skill", "creating a skill", or requests to turn a workflow, domain expertise, or tool into a reusable agent skill. Also use when the user wants to convert existing documentation, notes, or conversation history into a structured skill, or when they want to evaluate or refactor a skill they already have.

Particle Swarm Optimization (PSO) skill. A gradient-free, population-based metaheuristic for continuous function minimization, hyperparameter tuning, and black-box problems. Covers velocity and position updates, inertia weight, cognitive and social coefficients, constriction factor, stability and convergence analysis, worked examples on the sphere function and linear regression, and comparisons with gradient descent, least squares, and Bayesian optimization. Use when you need swarm intelligence, derivative-free optimization, or a simple global search over a continuous space.

Polars skill — Arrow-backed DataFrame library. Covers expression API, lazy/eager evaluation, group_by, joins, window functions, CSV/Parquet I/O, and pandas migration for high-performance data pipelines.

Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.