发现 Skills

Performs CSV/Excel data exploration and profiling. Use this when you request things like "Analyze this data", "Summarize the CSV", "Find outliers", or "Check data quality". Generates column summaries, missing value and outlier detection, correlation analysis, and data quality reports.

[Responsibility boundary] Dedicated to authentication, environment variables, and troubleshooting for the four connectors Drive · Notion · Higgsfield · OpenAI. Clearly separated from the pair moai-commerce:commerce-morning-brief (MCP store data retrieval) — this skill covers connector installation and authentication steps, while the pair is responsible for the actual MCP call results after authentication. Always use this skill for requests such as: "MCP connector connection", "Drive authentication method", "Where to get Notion Integration Token", "How to issue a Higgsfield key", "How to issue an OpenAI API key", "Windows MAX_PATH error", "Korean filename 30-character error", "computer:// link won't open", "How to connect 4 connectors", "MCP 4-connector authentication". v2.3.0 new (Wave 3 — Day 1 S4 setup).

[Responsibility boundary] Day 3 S4 dedicated to automatic model routing for advertising videos. Backend integration: Kling 3 (direct Higgsfield MCP) + Veo 3 · Seedance 2.0 (delegated via fal-gateway). Clearly distinguished from paired skills video-gen (general video generation) · image-gen (image-only) — differentiation keys are category-matrix-based automatic routing + suspicion-blocking hook pattern + generation of main/auxiliary video sets. Use this skill for requests such as: "Make an ad video", "main cut video", "automatic ad video generation", "Day3 S4 video", "suspicion-blocking hook video", "ecommerce ad video", "auxiliary 2-cut video", "Kling·Veo·Seedance routing". V6 Day3: ⑮ one main video 5–10 seconds + ⑯ two auxiliary cuts (Day3 S4 14:08–14:20 instructor demo, SPEC §3.1). [Internal materials] references/category-matrix.md: category matrix for instructor/development-team internal use. Do not expose to students (REQ-MEDIA-013). references/hook-pattern.md: analysis of the suspicion-blocking hook (internal).

Set up Claude Code hooks to block dangerous git commands (push, reset --hard, clean, branch -D, etc.) before they execute. Use when user wants to prevent destructive git operations, add git safety hooks, or block git push/reset in Claude Code.

Bytecode interpreter and JIT compiler skill for implementing language runtimes in C/C++. Use when designing bytecode dispatch loops (switch, computed goto, threaded code), implementing stack-based or register-based VMs, adding a simple JIT using mmap/mprotect, or understanding performance trade-offs in interpreter design. Activates on queries about bytecode VMs, dispatch loops, computed goto, JIT compilation basics, tracing JITs, or implementing a scripting language runtime.

Rust compiler skill for systems programming. Use when selecting RUSTFLAGS, configuring Cargo profiles, tuning release builds, reading assembly or MIR output, understanding monomorphization, or diagnosing compilation errors. Activates on queries about rustc flags, Cargo.toml profiles, opt-level, LTO, codegen-units, target-cpu, emit asm, MIR, or Rust build performance.

Toolkit for styling artifacts with a theme. These artifacts can be slides, docs, reportings, HTML landing pages, etc. There are 10 pre-set themes with colors/fonts that you can apply to any artifact that has been creating, or can generate a new theme on-the-fly.

/home/moli/.claude/skills/gstack/careful/SKILL.md

ClickHouse database patterns, query optimization, analytics, and data engineering best practices for high-performance analytical workloads.

Converted from Claude plugin command "insights" (PostHog). Use when the user wants to run this slash-command workflow. Query PostHog analytics and insights.

Identify user needs for Wardley Mapping from prior research. Reads shared research and the project brief, proposes users and their needs, then facilitates negotiation with the client. Produces a markdown needs document that must be explicitly agreed before proceeding. Use after wm-research is complete and brief.agreed.md exists.

Automatically activates when a dbt model, SQL file, or table is referenced. Surfaces Monte Carlo context — table health, active alerts, lineage, blast radius — before any code is written, and uses those findings to shape code recommendations. Generates and optionally deploys monitors for new transformation logic. After a change is made, generates targeted SQL validation queries to verify the change behaved as intended. Do not wait to be asked: run the appropriate workflow as soon as a relevant file or table is referenced.

Backend API design, database architecture, microservices patterns, and test-driven development. Use for designing APIs, database schemas, or backend system architecture.

Analyze and optimize tweets for maximum reach using Twitter's open-source algorithm insights. Rewrite and edit user tweets to improve engagement and visibility based on how the recommendation system ranks content.

Land and deploy workflow. Merges the PR, waits for CI and deploy, verifies production health via canary checks. Takes over after /ship creates the PR. Use when: "merge", "land", "deploy", "merge and verify", "land it", "ship it to production". (gstack)

Use this skill when gathering requirements for new BigQuery data models OR when asked to edit existing queries in bqetl. For new models, it guides a structured requirements interview. For existing queries, it analyzes the current model, checks downstream dependencies, and gathers requirements for changes. Works as pre-planning before the query-writer skill.

ALWAYS use this skill when users ask to "create a skill", "make a skill for...", "add a new skill", or similar requests. This skill guides the creation of effective skills in the bigquery-etl repository that extend Claude's capabilities with specialized knowledge for BigQuery ETL workflows, Mozilla data platform conventions, or telemetry analysis. CRITICAL - First checks for conflicts with existing skills and recommends using/updating existing skills when appropriate. Do NOT attempt to create skills without invoking this skill first.

Manage dialogue and text content so it remains editable, implementable, localizable, and testable.

Audit whether the local BSPDN physical contract is coherent across paper assumptions, GT3 tech collateral, layer/via topology, and current flow policy before promotion or expensive attribution experiments.

Voice development and style coaching for creative writing. Analyzes the user's natural writing patterns, builds a voice profile, coaches on register transitions, and flags unintentional voice shifts. The voice is the user's — this skill helps them find and refine it.

Use when building Android UIs with Jetpack Compose, managing state with remember/mutableStateOf, or implementing declarative UI patterns.

Verify, commit, and finish work. Handles merge, PR creation, cleanup. Replaces branch-finisher and worktree-manager.

Define a unique brand identity per app — colors, typography, motion personality, haptic vocabulary. Use during new app setup and design refresh.

Creates distinctive, production-grade UI for HTML/CSS/JS and iOS/SwiftUI. Use when asked to design, build, or improve interfaces, screens, components, or pages. Commits to a bold aesthetic direction before writing code — no generic AI aesthetics.