The OpenClaw ecosystem has over 3,500 skills. That number keeps growing. If you’ve already set up your agent, the next question is obvious: which ones are actually worth installing?
We tested hundreds to find the best OpenClaw skills for 2026. Most are fine. Some are abandoned. A handful are good enough that they changed how we use our agents day to day.
This is that handful.
How We Picked These
Three criteria:
-
It works reliably. No broken installs, no missing dependencies, no silent failures. If it takes more than two minutes to set up, it didn’t make the list.
-
Someone maintains it. Active repositories with recent commits. Abandoned skills rot fast when APIs change.
-
It does something you’d actually use. A skill that converts CSV to JSON is technically functional. It’s not making this list.
We reviewed 433 curated skills from the full 3,500+ ecosystem. These are the ones we’d install on our own machines.
Editor’s Choice
Before the category breakdowns, four skills from the Trellis team that we use daily.
| Skill | What It Does |
|---|---|
| Trellis | AI workflow system for Claude Code and Cursor — structured specs, skills, and automation |
| trellis-meta | Meta-skill for understanding and customizing Trellis workflows |
| Spec Templates | Ready-to-use spec templates for Electron apps, APIs, and more |
| Open Typeless | Real-world use case: AI-powered typing practice app built with Trellis |
These are opinionated picks. Trellis is a workflow system — if you’re using Claude Code or Cursor for development, it adds structure to how your AI agent works. If you’re not a developer, skip ahead to the category picks.
Best Media Skills
Media skills handle images, video, audio, and design. This is the biggest category and the one where skill quality varies the most.
fal-ai
Image, video, and audio generation through the fal.ai API. Supports FLUX, SDXL, Whisper, and a dozen other models. One skill, many backends.
What makes it stand out: you get access to multiple generation models without installing separate skills for each one. Ask for an image and it routes to the right model. The API is fast and the results are consistent.
Needs: fal.ai API key
elevenlabs-skill
Text-to-speech, sound effects, music generation, and voice management. If you need your agent to produce audio, this is the one to install.
The voice cloning feature lets you create a consistent narrator voice and reuse it across outputs, which saves time if you’re producing content regularly.
Needs: ElevenLabs API key
figma
Design analysis and asset export from Figma files. Send your agent a Figma link and it extracts design data, component structures, and exportable assets.
Useful for developers who receive Figma designs and want to pull measurements, colors, or assets without opening the Figma app.
Needs: Figma access token
ffmpeg-video-editor
Describe what you want to do to a video in plain language, and it generates the FFmpeg command. Cut, trim, convert, compress, add subtitles — without memorizing FFmpeg’s syntax.
Anyone who’s spent twenty minutes on Stack Overflow trying to figure out the right FFmpeg flags will appreciate this one.
tube-summary
Search YouTube for videos on any topic and get summaries from the subtitles. Useful for research — find the key points in a 45-minute talk without watching the whole thing.
Works best with videos that have good subtitles. Auto-generated captions produce noisier results.
Full category: All Media Skills
Best Productivity Skills
Productivity skills connect your agent to the tools you already use for work.
google-workspace-mcp
Gmail, Calendar, Drive, Docs, and Sheets, all through one skill. The setup is the selling point: OAuth sign-in, no Google Cloud Console configuration. You sign in and it works.
Most Google integration skills require you to create a GCP project, enable APIs, generate credentials, and configure OAuth consent screens. This one skips all of that.
gamma
Generate presentations, documents, and social posts using the Gamma.app API. Describe what you want, and it builds a formatted deck.
Good for quick drafts when you need a presentation in minutes rather than hours. You’ll still want to edit the output, but it gets the structure and layout right.
Needs: Gamma.app account
clinkding
Bookmark management through linkding. Save URLs, search your bookmarks, tag and organize. If you self-host linkding, this skill turns your agent into a bookmark assistant.
Niche, but if you’re a linkding user, it’s exactly what you want.
Needs: linkding instance URL and API token
Full category: All Productivity Skills
Best Development Skills
Development skills help with databases, deployment, code visualization, and infrastructure.
neondb-skill
Manage Neon serverless Postgres databases from your agent. Create projects, branches, databases, and run queries through conversation.
Neon’s branching model makes this skill especially useful — spin up a database branch, test a migration, drop it if it breaks. All without leaving your chat window.
Needs: Neon API key
azd-deployment
Deploy containerized applications to Azure Container Apps using the Azure Developer CLI. Handles the deployment pipeline through conversation instead of YAML files.
If you deploy to Azure and find the portal overwhelming, this cuts through the complexity.
Needs: Azure CLI configured
supabase-rls-gen
Generate Supabase Row Level Security policies from your Prisma schema. Feed it your schema, and it produces RLS policies that match your data model.
Writing RLS policies manually is tedious and error-prone. This does the mechanical work so you can focus on the access rules that actually matter.
Needs: Prisma schema file
diagram-gen
Generate Mermaid diagrams from your codebase. Point it at a directory and it produces class diagrams, sequence diagrams, and dependency graphs.
Useful for documentation and for understanding unfamiliar codebases. The diagrams aren’t perfect, but they give you a starting map.
Full category: All Development Skills
Best AI Tools
AI Tools skills give your agent access to other AI models and generation APIs.
pollinations
The Pollinations.ai API for text, image, video, audio generation, and analysis. Similar to fal-ai in scope but uses different backends.
The key difference: Pollinations offers some free-tier generation. If you want to experiment without committing to another API subscription, start here.
runware
Image and video generation through the Runware API. Access to FLUX, Stable Diffusion, and Kling AI models.
Where fal-ai gives you breadth across modalities (image, video, audio), Runware focuses on visual generation with more model options and fine-tuning controls.
Needs: Runware API key
nvidia-image-gen
Image generation and editing using NVIDIA’s FLUX models. Focused specifically on NVIDIA’s inference pipeline.
If you have NVIDIA hardware or want to use NVIDIA’s cloud inference, this is the direct path.
Needs: NVIDIA API key
Full category: All AI Tools
Best Automation Skills
Automation skills handle browser sessions, terminal management, and workflow orchestration.
vibetunnel
Manage terminal sessions from your agent. Create, list, monitor, and control terminal sessions remotely.
Think of it as giving your agent its own set of terminal windows. Useful for running background processes, monitoring logs, or managing multiple tasks without switching contexts.
browser-cash
Spin up browser sessions through Browser.cash for web automation. The sessions bypass common anti-bot protections, which matters for scraping, testing, and automation tasks that regular headless browsers can’t handle.
Needs: Browser.cash account
Full category: All Automation Skills
Best Smart Home Skills
Smart Home skills let your agent control physical devices through your messaging app.
samsung-smartthings
Control Samsung TVs and SmartThings-connected devices via the SmartThings API. Turn things on and off, change inputs, adjust settings. All through conversation.
The setup requires creating a SmartThings OAuth app, which takes about ten minutes. After that, you can control your TV from Telegram.
Needs: SmartThings OAuth app
sonoscli
Control Sonos speakers: discover devices, check status, play music, adjust volume, manage speaker groups. It does what you’d expect, and it does it well.
Tell your agent “play jazz in the living room at 30% volume” and it handles the rest.
Full category: All Smart Home Skills
What to Install First
Too many choices? Here’s what to install based on what you use your agent for.
”I just set up OpenClaw and want to try something cool”
clawhub install fal-ai
clawhub install tube-summary
Generate an image. Summarize a YouTube video. Two skills that show you what an AI agent can do beyond text chat.
”I use OpenClaw for work”
clawhub install google-workspace-mcp
clawhub install diagram-gen
clawhub install gamma
Email, calendar, and docs through your agent. Diagrams from your code. Presentations when you need them.
”I’m a developer”
clawhub install neondb-skill
clawhub install diagram-gen
clawhub install supabase-rls-gen
Database management, architecture visualization, and security policy generation. Developer workflow tools.
”I want to control my home”
clawhub install samsung-smartthings
clawhub install sonoscli
TV and speakers through your messaging app.
Installing Skills
Every skill on this list installs the same way:
clawhub install skill-name
Then verify it loaded:
openclaw skills list
If you haven’t set up OpenClaw yet, start with our Getting Started guide. It covers installation, API key configuration, and connecting your first messaging channel.
For the full curated list beyond this article, see Best Skills 2026. We update that page as new skills come out.
FAQ
How many skills should I install?
Start with two or three. Each skill adds context to your agent, which uses tokens. More skills means higher API costs per message. Add skills as you find specific needs, not all at once.
Do skills conflict with each other?
Rarely. Skills are independent modules. The main risk is two skills that handle similar tasks (like fal-ai and pollinations for image generation). Your agent might pick the wrong one. If that happens, uninstall the one you use less.
Are these skills free?
The skills themselves are free and open-source. Some require API keys from third-party services (fal.ai, ElevenLabs, Neon, etc.), which may have their own pricing. We noted which skills need external API keys in each section.
How do I uninstall a skill?
clawhub uninstall skill-name
Where can I find more skills?
- Claw Directory — 433+ curated skills with descriptions and categories
- Best Skills 2026 — Our continuously updated picks
- Run
clawhub search <keyword>to browse the full 3,500+ skill ecosystem from your terminal
Summary
| Category | Top Pick | Why |
|---|---|---|
| Media | fal-ai | Multi-model image/video/audio generation in one skill |
| Productivity | google-workspace-mcp | Gmail, Calendar, Drive without GCP setup |
| Development | neondb-skill | Serverless Postgres with branching, managed through chat |
| AI Tools | pollinations | Free-tier AI generation across multiple modalities |
| Automation | vibetunnel | Terminal session management from your agent |
| Smart Home | sonoscli | Speaker control that just works |
The full list of curated skills lives at Claw Directory. We update it regularly as new skills ship and old ones improve.