发现 Skills

Apply behavioral psychology and cognitive biases to marketing for conversion optimization, pricing strategy, copy improvement, and campaign design. Includes 70+ mental models organized by application, diagnostic frameworks for identifying behavioral barriers, and implementation guides with before/after examples. Use when applying psychology to marketing, optimizing conversions, improving pricing pages, or when user mentions psychology, cognitive bias, persuasion, behavioral science, mental models, why people buy, decision-making, consumer behavior, or neuromarketing.

Specialized for searching existing open-source libraries, tools, MCP servers, or best-practice code on GitHub. Use it when you want to check for an 'off-the-shelf' solution or reference examples before starting development.

Generate and maintain citation-backed repository wikis as MkDocs-compatible Markdown. Creates baseline docs, detects changes, and incrementally updates only affected pages with file path and line number citations.

Build personalized cold outreach sequences for LinkedIn and email. Use when someone needs to reach prospects, warm up cold leads, or build a systematic outreach engine. Covers research, connection requests, follow-ups, and conversion.

Pre-match alpha packets with price predictions, historical patterns, and optimal entry/exit timing for upcoming matches.

WeChat Official Account content writer with article formatting, headline optimization, and engagement tips. Use when you need to write WeChat articles, optimize headlines for WeChat, format content for mobile reading, or plan WeChat content strategy. Triggers on: WeChat Official Account, WeChat, article writer, formatting, headline, title.

Game art principles. Visual style selection, asset pipeline, animation workflow.

Best practices for Remotion - Video creation in React

Edit videos locally using ffmpeg. Trim, concat, resize, speed, overlay, extract audio, compress, and convert. Use when: (1) Trimming or cutting video segments, (2) Concatenating multiple clips, (3) Resizing video for social platforms, (4) Extracting or replacing audio, (5) Compressing video, (6) Converting video formats, (7) Getting video info.

GitHub patterns using the gh CLI for pull requests, stacked PRs, code review, branching strategies, and repository automation. Use when working with GitHub PRs, merging strategies, or repository management tasks.

Use when preparing a Bknd application for production deployment. Covers security hardening, environment configuration, isProduction flag, JWT settings, Guard enablement, CORS, media storage, and production checklist.

Cursor PM でアイデアを検証し Plans.md を更新してバトンタッチ。Cursor ↔ Claude Code 2-Agent ワークフロー対応。Use when user mentions Cursor PM handoff, 2-agent plan validation, CC-Cursor round trip, or brainstorm review. Do NOT load for: implementation work, single-agent tasks, or direct coding.

Postgres performance optimization and best practices from Supabase. Use this skill when writing, reviewing, or optimizing Postgres queries, schema designs, or database configurations.

Use when implementing or debugging ANY network connection, API call, or socket. Covers URLSession, Network.framework, NetworkConnection, deprecated APIs, connection diagnostics, structured concurrency networking.

Use when ANY beta tester reports a crash, ANY crash appears in Organizer or App Store Connect, crash logs need symbolication, app was killed without crash report, or you need to triage TestFlight feedback

Uses Bazel to run builds, tests and linting

Generate context prompts for AI-to-AI handoffs. Use when sharing conversation context, continuing work in a fresh session, or handing off to another LLM. Supports full, brief, debug, architect, diff, general, research, career, and learning modes. Use for "cp", "complete prompt", "context prompt", "handoff prompt", "share context", or "hand off to another AI".

Run ClickHouse queries for analytics, metrics analysis, and event data exploration. Use when you need to query ClickHouse directly, analyze metrics, inspect event tracking data, or test query performance. Read-only by default.

Meta's 7–8B specialized moderation model for LLM input/output filtering. Six safety categories — Violence/Hate, Sexual Content, Weapons, Regulated Substances, Suicide/Self‑Harm, Criminal Planning. Reported accuracy: 94–95%. Deployable via vLLM, Hugging Face, and SageMaker. Integrates with NeMo Guardrails.

Give your AI agent eyes to see the entire internet (15+ platforms). Search and read Twitter/X, Reddit, YouTube, GitHub, Bilibili, Xiaohongshu, Douyin, Weibo, WeChat Official Accounts, LinkedIn, V2EX, RSS, Exa web search, and any web page. Use when the user asks to search or read any of these platforms, shares a URL, searches the web, or researches a topic. Triggers: 'search twitter', 'search x', 'search reddit', 'search youtube', 'youtube transcript', 'search github', 'search bilibili', 'search xiaohongshu', 'search douyin', 'search weibo', 'search wechat article', 'read this link', 'open this link', 'watch video', 'search online', 'web search', 'find information', 'research', 'help me check', 'help me configure', 'configure', 'help me install', 'search v2ex', 'v2ex', 'github issue', 'read tweet', 'Bilibili', 'Douyin', 'RSS', 'LinkedIn', 'Exa web search'.

Three.js asset loading - GLTF, textures, images, models, async patterns. Use when loading 3D models, textures, HDR environments, or managing loading progress.

Reviews Workers and Cloudflare Developer Platform code for type correctness, API usage, and configuration validity. Load when reviewing TypeScript/JavaScript using Workers APIs, wrangler.jsonc/toml config, or Cloudflare bindings (KV, R2, D1, Durable Objects, Queues, Vectorize, AI, Hyperdrive).

Diagnose or 排查 why a `/rag-recall/api/search/keyword` request did not recall an expected doc or FAQ in sit, uat, or prod. Use when the user wants to 排查 keyword 检索漏召回, provides a full keyword request or an existing requestId plus target docId/faqId values, and expects end-to-end troubleshooting through a Playwright-attached browser session, `trace/recordInfo`, ELK `traceTargetIds` logs, and ES ranking or `_explain` checks. Also use when the user asks to fabricate a live keyword request with `traceTargetIds` and then investigate the resulting requestId.

Use when the task requires automating a real browser. This skill is MCP-only and uses `playwright-ext` (`@playwright/mcp --extension`) to attach to the browser extension session.