Explain concepts through detailed, interactive machines drawn from an angled view. The skill produces standalone HTML, SVG, or React components, with moving parts and numerical readouts driven by the same model so readers can explore how the idea works.
Turn audio and video into text across 99 languages, translate speech into English, and create timestamped subtitles. The skill covers model selection, transcription settings, and batch processing for recordings such as meetings and podcasts.
Access paid APIs using Machine Payments Protocol (MPP), a protocol for request payments. The skill checks payment requirements, selects a compatible wallet or client, submits payment, and helps inspect the API response and receipt.
Helps you use Unsloth for language model fine-tuning with LoRA and QLoRA, methods that adapt models with fewer trainable parameters. Supports API questions, implementation, and debugging using reference material drawn from official documentation.
Use the hf command-line tool to find, download, and upload models and datasets on Hugging Face Hub. The guide also covers repository management, hosted computing resources, local cache maintenance, and automation options.
Set up vLLM to serve large language models through an OpenAI-compatible API or process prompts in batches. The skill covers deployment, performance monitoring, and quantization, which reduces model memory requirements.
Get guidance for fine-tuning large language models with Axolotl, adapting them through additional training. Use YAML configuration examples, dataset references, and training methods such as LoRA to set up training and investigate problems.
Generate model responses in defined formats for data extraction and classification. Use JSON schemas, Pydantic data models, regular expressions, or fixed choices to produce structured results that other programs can parse and validate.
Find language models on Hugging Face Hub and run them locally with llama.cpp. The skill helps compare GGUF model files, choose compressed versions that fit available memory, and prepare commands for a local chat server or command-line session.
Use SAM to separate objects from images by marking points or drawing boxes, without training a model for each task. Covers automatic region generation, refining selections, preparing annotation data, and exporting models for deployment.
Build question answering and document retrieval workflows with DSPy by defining inputs, outputs, and reusable components. Use training examples and evaluation metrics to automatically optimize prompts and compare results before and after optimization.
Guide language model training with TRL using example answers, preferred response pairs, or reward scores. The skill covers data preparation, method selection, reward model training, evaluation, and adjustments for memory limits or unstable training.
Record a website's network traffic in a HAR file, a log of requests and responses, then derive an HTTP client from its JSON requests. Reuse captured request details to run searches or fetch data without repeated browser interaction.
Carry out an implementation plan by assigning each task to a fresh subagent, a separate agent with its own context. Review specification compliance first, then code quality, repeat fixes as needed, and finish with an integration review.
Search across multiple engines through SearXNG, a service that combines search results. Filter by category, date range, or engine, and get titles, links, and snippets. Public instances require no API key.
Read Reddit posts, comments, and user activity without a browser or account. Use it to find discussions, summarize threads, and explore community reactions, with optional app credentials for more detailed data and frequent requests.
Set up and optimize large language model inference, or response generation, on NVIDIA GPUs. Use quantization, batch processing, and multiple GPUs to improve serving performance, and expose models through a chat completion API.
Build agent memory with Pinecone by storing searchable representations of conversations and documents. Retrieve relevant past context for answers, organize memory by session or user, and prototype semantic search that finds information by meaning.
Publish a Cloudflare Worker to a temporary live URL without creating an account first. The skill covers deployment, checking the response, updating the code, and handing over an ownership link before the 60-minute expiry.
Set up services such as databases and SMS for a project through Stripe Projects. Sync generated credentials into `.env`, connect supported provider accounts, and manage credential rotation, plan changes, and service removal.
Support drug discovery research by searching public databases for compound activity, molecular properties, drug interactions, and gene–disease associations. Assess drug-likeness, meaning how molecular properties fit established rules for potential oral medicines.
Cross-reference public records to investigate companies, contracts, sanctions, property, and litigation. Match names across sources, label confidence levels, and build structured findings linked to original records, with optional statistical analysis of event timing.
Publish static websites and built web apps to GitHub Pages, Cloudflare Pages, or Netlify. The workflow includes preview approval, git-tagged deployment versions, live URL checks, and restoring an earlier version when an update needs reversing.
Add an AI assistant to an existing web app so users can operate its interface with everyday instructions. Covers embedding the panel, connecting a language model, testing interactions, and protecting API keys in production.
Is this your account? Link GitHub to bring it onto your creator page (Japanese)