Claude Skills · Weekly
15 skills across 5 tracks.
Clone voices, dub videos into 646 languages and generate audiobook narration entirely on your own machine, with no per-minute API bill. Built for podcasters and YouTubers who want ElevenLabs-grade voice work without shipping their audio to a cloud service.
⭐ 42.5k stars · debpalash
To install, run this command in Claude Code:
Turn a long recording into ranked short clips, reframe them to 9:16 and cut a finished edit straight from the transcript by prompting your agent. For creators who want Opus Clip and CapCut results as a scriptable open-source pipeline they actually control.
⭐ 13 stars · WhiteTowerAI
To install, run this command in Claude Code:
Transcribe multi-speaker podcast and interview audio across 99 languages with speaker diarization and an LLM cleanup pass, on GPU or CPU. Ideal for interview-format creators who need clean transcripts for show notes, subtitles and repurposing.
⭐ 11 stars · zxkane
To install, run this command in Claude Code:
Build, debug and harden containerized apps with Docker's own workflows for multi-stage Dockerfiles, Compose stacks and sandboxed agent execution. For full-stack developers who want their agent writing container config that survives the jump from local dev to production.
⭐ 391 stars · docker
To install, run this command in Claude Code:
Wire up serverless Redis caching, HTTP message queues, cron schedules and durable multi-step workflows with per-SDK instructions for Redis, QStash, Workflow, Vector and Ratelimit. For backend and edge developers on serverless runtimes where connection pooling and long-running jobs are the hard part.
⭐ 27 stars · upstash
To install, run this command in Claude Code:
Instrument services with vendor-neutral OpenTelemetry across Node.js, Go, Python, Java, .NET, Ruby and PHP, then get the semantic-convention attributes, Collector config and OTTL transforms right the first time. For backend engineers adding tracing and metrics who do not want their telemetry locked to one vendor's agent.
⭐ 94 stars · dash0hq
To install, run this command in Claude Code:
Query GenAI traces, token spend and agent failures in DQL, wire up OneAgent instrumentation, and generate dashboards or notebooks straight from the terminal. For AI engineers running LLM features in production who need root cause, not just a latency chart.
⭐ 161 stars · Dynatrace
To install, run this command in Claude Code:
Build BM25, semantic, hybrid and agentic search on OpenSearch from an empty index, then pivot to querying logs and distributed traces with PPL in the same session. For AI engineers whose RAG retrieval layer and agent telemetry both live in a search cluster.
⭐ 55 stars · opensearch-project
To install, run this command in Claude Code:
Declare multimodal tables in Python that hold images, video, audio and documents next to computed columns, embedding indexes and UDF-backed model calls, then serve them over FastAPI routes. For AI engineers who want one store for media, embeddings and inference instead of gluing a vector DB to a job queue.
⭐ 5 stars · pixeltable
To install, run this command in Claude Code:
Audit any interface against Apple's Human Interface Guidelines — app icons, navigation structure, dark mode, Liquid Glass, accessibility — and get back specific cited fixes instead of vague vibes. For designers and design engineers shipping cross-platform UI in SwiftUI, Flutter, Tauri or React Native who want a reviewer that knows the actual rules.
⭐ 847 stars · dickwu
To install, run this command in Claude Code:
Audit your React components for real-versus-slop quality, capture the flows that matter, and write CSF3 stories that document behavior rather than just screenshotting it. For design engineers maintaining a component library who want Storybook to stay honest as the system grows.
⭐ 36 stars · strongeron
To install, run this command in Claude Code:
Build page transitions, micro-interactions and SVG or Lottie motion that hold 60fps and respect prefers-reduced-motion, with GSAP patterns the agent applies instead of improvising. For UI designers who want motion that feels intentional rather than a pile of ad-hoc CSS keyframes.
⭐ 28 stars · iart-ai
To install, run this command in Claude Code:
Audit landing pages, spec build-ready sales funnels and wire up server-side conversion tracking that survives iOS signal loss, using six composable playbooks. For growth marketers who own funnel performance end to end instead of only the ad spend.
⭐ 256 stars · autonnel
To install, run this command in Claude Code:
Design, launch, ramp and read out A/B tests and feature-flag rollouts against your GrowthBook instance across 25+ terminal workflows for experiments, flags, metrics and dashboards. For growth and RevOps teams running an experimentation program who want the same stats discipline applied every time.
⭐ 23 stars · growthbook
To install, run this command in Claude Code:
Query GA4 custom reports, realtime active users and custom metric definitions as structured JSON through 30+ commands built on Google's official Admin and Data APIs. For marketers and analysts who would rather script recurring reporting than rebuild the same GA4 exploration by hand each month.
⭐ 39 stars · Bin-Huang
To install, run this command in Claude Code:
Want these in your inbox?
Three curated Claude Code skills, every Monday. Pick your tracks.
Subscribe free