Skill UI

Guides teams through creating interactive, AI-enhanced OpenAPI/AsyncAPI docs, developer portals, SDKs, and onboarding workflows plus automated testing and localization to speed integrations and maintain accuracy.

Activation-aware weight quantization that compresses 7B-70B models to 4-bit with minimal accuracy loss, delivering ~3x inference speedups and memory reductions for constrained GPU deployments while integrating with vLLM and Marlin kernels.

Guides users through writing, reviewing, and editing documentation with consistent structure, tone, and accuracy; ideal when creating README files, updating docs folders, or refining markdown drafts per contribution standards.

Feature Engineering Toolkit automates creating, selecting, and transforming dataset features to improve machine learning models; use it whenever you need new interactions, automatic selection, scaling, or encoding to boost accuracy and interpretability.

Provides detailed evaluation of machine learning models using plugin metrics to assess accuracy, precision, recall, F1, and other key indicators, helping users compare models and validate performance before deployment.

FAISS (Facebook AI Similarity Search) is a library designed for efficient similarity search and clustering of dense vectors. It supports indexing and searching on billions of vectors at scale, featuring GPU acceleration and multiple index types (Flat, IVF, HNSW, PQ) to balance speed and accuracy. It is essential for building high-performance, large-scale applications requiring fast k-NN retrieval or pure vector similarity matching, especially in fields like RAG and recommendation systems.

Guides optimizing Granola transcription accuracy and note quality through audio setup, meeting practices, template design, post-meeting review, and integration tuning.

This skill guides users through advanced prompt optimization techniques for Large Language Models (LLMs). It functions as an expert prompt engineer coach, applying methodologies like Constitutional AI and Chain-of-Thought reasoning. Use it to transform basic, raw instructions into highly effective, production-ready prompts, significantly improving accuracy, reducing hallucinations, and optimizing token usage for cost efficiency.

Optimizes deep learning models by analyzing architecture, data, and metrics to select optimizers, learning rate schedules, and regularization strategies, then outputs optimized code and evaluates accuracy, time, and resource usage.

Quantizes HuggingFace LLMs to 8-bit or 4-bit with bitsandbytes, cutting memory by up to 75% while keeping accuracy, and supports QLoRA fine-tuning plus 8-bit optimizers for faster, memory-efficient training.

Offers a decision framework that starts with regex for structured text extraction, flags low-confidence items, and only triggers a cheap LLM validator for edge cases to balance cost and accuracy.

Centralizes pipeline health checks, forecast accuracy tracking, and go-to-market efficiency measurement for SaaS revenue teams, generating text or JSON outputs via CLI scripts to monitor coverage, MAPE, GTM metrics, and risk flags before leadership reviews.