Browse reusable Agent Skills, each with a clear purpose and practical guidance.
Use when formulating the best technical approach for a task — before writing implementation plans or code. Triggers on architecture decisions, complex features, refactors, or when the user asks how to approach something. Investigates current codebase, questions existing patterns, researches alternatives, and presents approaches with trade-offs.
Use when you want to iterate on a feature interactively with the user watching a running instance — pair-programming rhythm for UI-heavy work, redesigns, copy tightening, layout exploration, or any "let me just show you what I want" session. Not for systematic audit (/live-exercise) or perf-tuning (/live-perf).
Use when touching any CSS — new files, edits, audits, refactors, `<style>` blocks, CSS-in-JS, Tailwind config, design-token files, or reviewing CSS a teammate wrote. Applies to vanilla CSS, shadow DOM internals, web-component consumers, VS Code webviews, and framework-scoped styles. Skip only for files with no styles at all (e.g. `.py`/`.go`/`.rs` without embedded CSS).
ET Luban export workflow. Use when exporting generated C# config/data code, refreshing aggregated luban.conf, validating CodeMode outputs, or troubleshooting ET.ExcelExporter, LubanGen.ps1, table definitions, beans, enums, Defines, and Luban export failures.
Pick the right fine-tuning framework for a MiniCPM5-1B base checkpoint and route to a framework-specific cookbook skill. Use when the user wants to SFT / LoRA / DPO / continue-pretrain MiniCPM5 and has not yet committed to a specific framework, or when they say "fine-tune MiniCPM5", "train MiniCPM5", "MiniCPM5 微调", "LoRA MiniCPM5", "继续训练 MiniCPM5".
Fine-tune MiniCPM5-1B into a LoRA adapter and convert it to a GGUF adapter that loads directly into llama.cpp / llama-server and the MiniCPM Desk Pet app's custom-LoRA upload. Use when the user wants "GGUF LoRA", "convert LoRA to GGUF", "convert_lora_to_gguf", a custom persona/skin for the desktop pet, "桌宠自定义 LoRA", "上传 LoRA 到桌宠", or asks how to take a trained adapter and run it on a GGUF base.
Fine-tune MiniCPM5-1B with LLaMA-Factory (YAML-driven SFT / DPO / WebUI). Use when the user wants to fine-tune via LLaMA-Factory, llamafactory-cli, mentions YAML configs, WebUI, or asks for the most-documented community framework.
Fine-tune MiniCPM5-1B with bare-metal TRL + PEFT, including assistant-only loss via a chat-template patch. Use when the user wants minimal Python, no YAML, full control, or asks for "TRL", "SFTTrainer", "PEFT", "LoraConfig", "assistant_only_loss".
Fine-tune MiniCPM5-1B with unsloth for tight-VRAM single-GPU LoRA / QLoRA. Use when the user wants "unsloth", "FastLanguageModel", QLoRA on a 24 GB consumer GPU, or asks for the smallest VRAM footprint.
Fine-tune MiniCPM5-1B with xtuner (mmengine config-driven SFT). Use when the user mentions "xtuner", "mmengine", InternLM's training framework, or wants config-file-driven training.