Loading...
Loading...
Found 365 Skills
Compare two datasets to find differences, added/removed rows, changed values. Use for data validation, ETL verification, or tracking changes.
Interact with Langfuse and access its documentation. Use when needing to (1) query or modify Langfuse data programmatically via the CLI — traces, prompts, datasets, scores, sessions, and any other API resource, (2) look up Langfuse documentation, concepts, integration guides, or SDK usage, or (3) understand how any Langfuse feature works. This skill covers CLI-based API access (via npx) and multiple documentation retrieval methods.
Use ONLY when creating NEW registrable components in ML projects that require Factory/Registry patterns. ✅ USE when: - Creating a new Dataset class (needs @register_dataset) - Creating a new Model class (needs @register_model) - Creating a new module directory with __init__.py factory - Initializing a new ML project structure from scratch - Adding new component types (Augmentation, CollateFunction, Metrics) ❌ DO NOT USE when: - Modifying existing functions or methods - Fixing bugs in existing code - Adding helper functions or utilities - Refactoring without adding new registrable components - Simple code changes to a single file - Modifying configuration files - Reading or understanding existing code Key indicator: Does the task require @register_* decorator or Factory pattern? If no, skip this skill.
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.
Use when the user needs Excel file manipulation — reading, writing, formulas, charts, conditional formatting, data validation, pivot tables, or large file handling. Trigger conditions: create Excel reports programmatically, read spreadsheet data, add formulas or charts, apply conditional formatting, perform data validation, generate pivot tables, handle CSV import/export, process large datasets in Excel format.
Roll out self-serve analytics on MotherDuck for internal teams. Use when deciding the first governed dataset, the first Dive or share, ownership boundaries, and the rollout path from one audience to broader adoption.
Use this skill when you have structured course content (or any chapter-based dataset) in markdown form and need to turn it into a working interactive website — without picking a framework, without a build step. Triggers on phrases like "做成網頁", "轉成 SPA", "course-data.js", "render 函式", "把講義變網頁", "static site from markdown", "vanilla JS site", "no-framework site", "single-page app from markdown". The output is a vanilla HTML + JS single-page app that opens with `npx serve` and persists state in localStorage. Always invoke AFTER `course-content-authoring` (content stable), BEFORE `static-spa-interactions` (this skill produces the scaffold; interactions adorn it).
Use bigquery CLI (instead of `bq`) for all Google BigQuery and GCP data warehouse operations including SQL query execution, data ingestion (streaming insert, bulk load, JSONL/CSV/Parquet), data extraction/export, dataset/table/view management, external tables, schema operations, query templates, cost estimation with dry-run, authentication with gcloud, data pipelines, ETL workflows, and MCP/LSP server integration for AI-assisted querying and editor support. Modern Rust-based replacement for the Python `bq` CLI with faster startup, better cost awareness, and streaming support. Handles both small-scale streaming inserts (<1000 rows) and large-scale bulk loading (>10MB files), with support for Cloud Storage integration.
Create diverse synthetic test inputs for LLM pipeline evaluation using dimension-based tuple generation. Use when bootstrapping an eval dataset, when real user data is sparse, or when stress-testing specific failure hypotheses. Do NOT use when you already have 100+ representative real traces (use stratified sampling instead), or when the task is collecting production logs.
Browser automation CLI with Nstbrowser integration for AI agents. Use when the user needs advanced browser fingerprinting, profile management, proxy configuration, batch operations on multiple browser profiles, or cursor-based pagination for large datasets. Triggers include requests to "use NST profile", "configure proxy for profile", "manage browser profiles", "batch update profiles", "start multiple browsers", "list profiles with pagination", or any task requiring Nstbrowser's anti-detection features.
Comprehensive guide for AntV L7 geospatial visualization library. Use when users need to: (1) Create interactive maps with WebGL rendering (2) Visualize geographic data (points, lines, polygons, heatmaps) (3) Build location-based data dashboards (4) Add map layers, interactions, or animations (5) Process and display GeoJSON, CSV, or other spatial data (6) Integrate maps with AMap (GaodeMap), Mapbox, Maplibre, or standalone L7 Map (7) Optimize performance for large-scale geographic datasets
Trains and fine-tunes vision models for object detection (D-FINE, RT-DETR v2, DETR, YOLOS), image classification (timm models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3 — plus any Transformers classifier), and SAM/SAM2 segmentation using Hugging Face Transformers on Hugging Face Jobs cloud GPUs. Covers COCO-format dataset preparation, Albumentations augmentation, mAP/mAR evaluation, accuracy metrics, SAM segmentation with bbox/point prompts, DiceCE loss, hardware selection, cost estimation, Trackio monitoring, and Hub persistence. Use when users mention training object detection, image classification, SAM, SAM2, segmentation, image matting, DETR, D-FINE, RT-DETR, ViT, timm, MobileNet, ResNet, bounding box models, or fine-tuning vision models on Hugging Face Jobs.