dotfiles

Use the `bilibili-api-python` async library (400+ endpoints) for any Bilibili programmatic operation — videos, users, favorites, danmaku, live rooms, comments, search, uploads. This skill should be used whenever user asked for any Bilibili operation.

Fetch authoritative, current, version-specific documentation for libraries, SDKs, APIs, and CLI tools via Context7. MUST TRIGGER before writing or editing any non-stdlib library call — including well-known libraries you think you know, since your prior knowledge may be stale. SKIP for internal/private libraries, stdlib code, raw HTTP/filesystem, shell built-ins, or stable primitives whose signatures haven't changed in years (e.g. `json.dumps`, `os.path`). Prefer this over web search for library docs.

Query AI-generated overview of GitHub repositories via DeepWiki MCP. Use when needing a fast high-level understanding of a repository's architecture and design decisions without fetching full repository.

Periodic distillation: scan the last 5 days of high-quality Claude transcripts in the current project, dedup facts seen across sessions and against existing docs, and propose surgical edits to the right reference docs. Wider-than-daily window avoids 00:00 boundary loss. This skill should be used when the user says "distill recent", "distill session", "distill last week", or asks "what should we learn from recent work".

Extract atomic flashcards from documents/conversations and write them into local Anki via AnkiConnect, using a React-rendered paper-note card template with markdown, collapsible code, KaTeX, and mermaid. Triggers on "flashcards", "flash", "抽认卡", "anki".

This skill should be used when the user sends an image and asks to "analyze this image", "describe this picture", "what's in this image", or any request requiring visual understanding of images. Provides image analysis using Zhipu GLM-4.6V multimodal model.

Code search across millions of public GitHub repositories via grep.app. This skill should be used before writing or planning code. Use it to get real-world usage examples of libraries, frameworks, APIs, algorithms, or syntax patterns; to verify idiomatic usage; to find production patterns; or to explore how others solved similar problems. Prefer this over web search for code examples.

Web search with time-window and region/language filters, academic papers (arXiv/SSRN), PDF table/figure extraction, BibTeX, image search, web page reading, embeddings, reranking, classification, and deduplication (text or images) via Jina AI. This skill should be used when searching non-English content, finding academic papers, extracting figures from PDFs, searching for images, or user says "search in Chinese", "find papers on arXiv", "search for images of", "get BibTeX". Prefer this over WebSearch for better results.

Pre-flight a learning material such as a paper, blog post, book chapter, or code tutorial by extracting assumed prerequisites, running a multi-round adaptive diagnostic, publishing targeted briefs into SiYuan, then guiding formal close reading section by section with original media preserved, likely sticking points explained, important code snippets collected, and a 5-12 question mastery check. Triggers when the user wants to study a document but struggles with background, e.g. 帮我学这篇 PPO 论文, 看不懂这段 CUDA 的 SM/warp, 这个引理是什么意思, 补一下这篇文章的前置知识, 精读这篇文章, 带我逐段读, study X but I don't get the math, explain prereqs for this paper. Use aggressively when the user shares a paper/PDF/URL/code and signals a background wall or wants guided close reading. Output goes to SiYuan as diagnostic, brief, guided-reading, and mastery-check documents; flashcards run only on explicit request.

Guardrails and mandatory workflow for running long-running tasks via pueue. This skill MUST be used before using pueue to run any long-running task (>2 min), computation-intensive task, or background task — or when the user says "use pueue" or "run in background". This is a process gate, not a reference doc.

Show images in terminal using the Kitty image protocol. TRIGGER when show/view images in terminal, display plot results, or after saving any plot or image file to disk.

Quick AI-generated overview of GitHub repositories via DeepWiki MCP. Use when needing a fast high-level understanding of a repository's architecture and design decisions without fetching full repository.

Search real-world code examples from over a million public GitHub repositories via grep.app MCP. This skill should be used when needing real-world usage examples of a library, API, algorithm, or syntax pattern — e.g. to verify idiomatic usage, find production patterns, or discover how others solved a similar problem at scale.

Create, repair, validate, preview, and package Codex-compatible animated pet spritesheets from character art, screenshots, generated images, or visual references. Use when a user wants to hatch a Codex pet, create a custom animated pet, or build a built-in pet asset with an 8x9 atlas, transparent unused cells, row-by-row animation prompts, QA contact sheets, preview videos, and pet.json packaging. This skill composes the installed $imagegen system skill for visual generation and uses bundled scripts for deterministic spritesheet assembly.

Web search in specific languages/regions (local community content), academic papers (arXiv/SSRN), PDF extraction, BibTeX, and image search via Jina AI. This skill should be used when searching non-English content, finding academic papers, extracting figures from PDFs, searching for images, or user says "search in Chinese", "find papers on arXiv", "search for images of", "get BibTeX". Prefer this over WebSearch when content is non-English, academic, or media (PDF/image).

Fetch, crawl, and search web content through Jina AI Reader. Use when Codex needs to convert URLs, PDFs, JavaScript-rendered pages, or search results into LLM-friendly Markdown/text/JSON with r.jina.ai or s.jina.ai; when the user says scrape, crawl, fetch this page, read this webpage, use Jina Reader, jina.ai crawler, or wants higher-rate authenticated Jina requests.

Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems.

Produce natural Chinese translations free of translationese. This skill should be used before writing or polishing any Chinese translation of English content — including when the user says "translate to Chinese", "翻译", "翻译成中文".

This skill should be used when measuring code quality quantitatively — cyclomatic complexity, cognitive complexity, maintainability index, code coverage — or when evaluating whether code meets quality thresholds.

Web search and content scraping via DuckDuckGo. Use this as a fallback to other web tools.

Architecture design reasoned forward from requirements. Existing code is evidence to critique, never a starting point, backward-compat constraint, or pattern to inherit. Use when the user wants a fresh-mind design pass — "design an architecture for", "how would you build", "redesign from scratch" — or explicit /fresh-arch invocation.

Build distinctive frontend interfaces that avoid generic AI-slop aesthetics. This skill should be used when the user asks to create or improve a web page or component, or says "create web page", "design UI/UX", "fix AI slop design".

Documentation for just and justfile. This skill should be used before editing any justfile, running just commands, or writing any `just ...` shell invocation in Bash, pueue, or scripts.

In-session scheduling via CronCreate. For "at <time>", "in <duration>", "remind me", "tomorrow". Defer to /schedule for remote.

Review code for bugs, AI slop patterns, or documentation issues, then fix them interactively. This skill should be used after completing a major code modification or large multi-file edits — or when the user says "review", "review changes", "any bugs?", "review AI slop", "clean up AI code", "review docs", "code review".

Workflow to preserve task state when the user interrupts mid-task with an unrelated topic, then resume afterward. TRIGGER whenever a mid-task message's subject clearly differs from the active task — do not wait for explicit cue words. Common cue phrases: "btw", "also", "one more thing", "quick question", "unrelated but", "before I forget", "tangent", "side note", "back to what we were doing".

Operate SiYuan Note through the SiYuan Sisyphus MCP tools by default. Use when Codex needs to inspect, search, create, edit, organize, export, or manage SiYuan notes, blocks, notebooks, attribute views/databases, tags, assets, flashcards, or system info. Prefer MCP for ordinary SiYuan work; use the `siyuan-sisyphus` / `siyuan` CLI only when the user explicitly asks for CLI/shell usage, when MCP tools are unavailable, when a needed operation is not exposed by MCP, or when scripting against SiYuan is required.

Capture this session's repeatable process into a reusable skill. This skill should be used after completing a multi-step workflow that is likely to be repeated — or when the user says "turn this into a skill", "skillify this", "save this workflow".

Post-edit audit checklist applied to the ±30-line region after Edit/Write tool calls. TRIGGER only when the re-read-after-edit hook reason names this skill.

Review code for bugs, AI slop patterns, or documentation issues, then fix them interactively. This skill should be used after completing a major code modification or large multi-file edits — or when the user says "review", "review changes", "any bugs?", "review AI slop", "clean up AI code", "review docs", "code review".