Awesome OpenClaw Skills: A Curated List of the Best Community and Official Skills

The official ClawHub registry lists 13,729 community-built skills as of February 28, 2026, and the largest curated index (VoltAgent) surfaces 5,400+ of them. Browsing either one is useful, but neither answers the question most power users bring: which skills should I install today, and when should I reach for each one?

This is our opinionated “awesome” list. Every entry passes three tests: it exists in the public ClawHub registry or an official SFAI Labs repository, it has had a commit in the last 90 days or is explicitly marked “Mature”, and it solves a problem we have hit in production. Entries are split into Official / SFAI Labs and Community, grouped by category, and graded by maintenance signal. Where a category is thin, we say so rather than padding.

How to Read This List

Each entry has four parts:

1. Name — the skill slug as published on ClawHub. This is what you install.
2. One-line description — what the skill does, in under 15 words.
3. When to use — a trigger phrase. If your current situation matches the trigger, reach for this skill.
4. Grade — a maintenance signal we assign based on last commit date and changelog activity.

Grading uses three labels:

Grade	Meaning
Active	Commit in the last 30 days, open issues being triaged
Mature	Stable, infrequent commits, no breaking changes in the last 6 months
Early	Less than 90 days old or fewer than 100 downloads — promising but unproven

We do not list Archived or Abandoned skills. If a skill has gone quiet for 180+ days with unanswered issues, it is cut from the list at the next review.

This list was last reviewed on April 14, 2026. We plan to revisit it quarterly. If a skill you rely on is missing, see Contributing to This List.

Official / SFAI Labs Skills

These skills are maintained by the OpenClaw core team or SFAI Labs engineers. They ship with the stock OpenClaw distribution or live in the openclaw/ organization on ClawHub. Treat them as the safe default.

openclaw/heartbeat-scheduler Controls the 30-minute autonomous execution loop. Configures cadence, quiet hours, and max-runs-per-day. When to use: you want the agent to act without being prompted and need guardrails on frequency. Grade: Active.

openclaw/memory-core The persistent memory layer. Stores conversation history, user preferences, and task state across runs. When to use: always installed. If it is not loaded, the agent treats every heartbeat as cold start. Grade: Active.

openclaw/telegram-bridge Handles Telegram message routing, command parsing, and multi-chat context. When to use: your primary interface is Telegram. Pair with openclaw/whatsapp-bridge or openclaw/slack-bridge if you need more than one channel. Grade: Active.

openclaw/mission-control The web dashboard for viewing skill runs, memory contents, and heartbeat logs. When to use: debugging why a skill ran or did not run, or reviewing what the agent did overnight. Grade: Active.

openclaw/skill-validator Lints a local SKILL.md file against ClawHub publishing requirements before you push. When to use: you are about to publish a custom skill and want to catch frontmatter errors early. See our skills development guide for context. Grade: Active.

openclaw/browser-mode Gives the agent a headless browser with cookie persistence for sites that require login. When to use: your workflow needs to read or act on pages behind authentication. Configure cautiously, see browser mode configuration. Grade: Active.

openclaw/audit-logger Writes every tool call, model response, and skill invocation to a tamper-evident log. When to use: compliance review, incident forensics, or any multi-user deployment. Pair with audit logging. Grade: Mature.

openclaw/model-router Routes requests to the cheapest model that can handle them, falls back to a premium model on retries. When to use: you want to cap spend without manually picking models for every task. Grade: Active.

Community Skills by Category

The community skills below have been independently verified against ClawHub. Install at your own risk and review the source before granting sensitive permissions. See How to Evaluate a Skill Before Installing.

Coding Agents and IDEs

This is the largest category in the ecosystem, with VoltAgent tracking 1,222 skills under this umbrella.

code-review-pr Generates a structured pull request review with inline suggestions and a risk summary. When to use: you want a second pass on a PR before asking a human reviewer. Grade: Active.

test-runner-unified A unified wrapper around Jest, PyTest, Mocha, Go test, and Cargo test that normalizes output into a single report. When to use: your repo has more than one test framework and you want a single command. Grade: Active.

git-flow-helper Handles branch creation, rebase conflicts, and commit message formatting with conventional commits. When to use: you commit dozens of times a day and want the boilerplate handled. Grade: Mature.

refactor-guardian Runs a pre-refactor safety check that maps call sites and identifies tests covering the target function. When to use: you are about to change a function signature and want a blast-radius report first. Grade: Active.

DevOps and Cloud

VoltAgent tracks 409 skills here. The signal-to-noise ratio is higher than in coding — most DevOps skills that get published get used.

terraform-planner Runs terraform plan, summarizes the diff in plain English, and flags destructive changes. When to use: you want a human-readable summary of a Terraform plan before applying. Grade: Active.

k8s-incident-responder Given a pod name or alert ID, pulls logs, recent events, and the last three deployments for context. When to use: an alert fires and you want the triage context auto-assembled. Grade: Active.

ci-pipeline-monitor Watches GitHub Actions or GitLab CI runs and posts a summary when a pipeline fails. When to use: you want push notifications when main breaks. Grade: Mature.

dockerfile-linter Lints Dockerfiles for layer-caching mistakes, missing health checks, and known base-image CVEs. When to use: any time you touch a Dockerfile. Grade: Active.

Web and Frontend

VoltAgent tracks 938 skills here. Most wrap a browser-automation or scraping library.

page-scraper-resilient Scrapes a page with automatic retry, proxy rotation, and CAPTCHA detection. When to use: your target site blocks naive scrapers. Pair with openclaw/browser-mode. Grade: Active.

accessibility-audit Runs axe-core against a URL and returns a prioritized list of WCAG violations. When to use: before shipping a public-facing page. Grade: Mature.

lighthouse-reporter Wraps Lighthouse and produces a diff against a previous run so you can see regressions. When to use: you want performance budgets enforced in CI. Grade: Active.

Security and Secrets

Smaller category. Five to eight skills are worth installing, the rest are wrappers around tools you probably already run.

secret-scanner Scans a repo or a pending commit for high-entropy strings, API key patterns, and leaked credentials. When to use: pre-commit hook, or after a merge from an external contributor. Grade: Active.

dependency-auditor Runs npm audit, pip-audit, or cargo audit depending on the project type and summarizes exploitable findings. When to use: weekly sweep, or when a new CVE hits the news. Grade: Active.

oauth-token-rotator Detects expiring OAuth tokens across connected services and prompts for re-auth before they expire. When to use: you have had integrations silently break because a token expired. See OAuth token expired. Grade: Mature.

Data and Research

A fast-growing category. Many skills here are thin wrappers around web search APIs, so apply the maintenance grade rule strictly.

lead-enricher Given a domain or an email, returns company size, stack signals, and recent hiring activity. When to use: inbound demo requests where you want context before the call. Grade: Active.

competitive-pricing-tracker Scrapes competitor pricing pages on a schedule and alerts on changes. When to use: your product competes in a transparent-pricing market. Grade: Mature.

research-briefer Given a topic, runs parallel Google, ChatGPT, and Claude queries and assembles a source-ranked brief. When to use: you are about to write a long-form article or prep for a meeting. Grade: Active.

Productivity and Communications

The category power users hit first. Quality varies widely, so we have included only skills we run daily.

meeting-notes-writer Listens to a meeting transcript and produces decisions, action items, and an owner per item. When to use: paste a transcript, get a structured summary. See meeting notes. Grade: Active.

email-triage Scans the inbox every heartbeat, archives marketing, drafts replies to known senders, and escalates the rest. When to use: you want inbox zero without reading every message. Grade: Active.

calendar-reshuffler Handles reschedule requests by finding mutually free slots and proposing three options. When to use: you reschedule more than two meetings a week. Grade: Mature.

standup-summary Reads the last 24 hours of commits, Slack messages, and ticket updates, and drafts your standup post. When to use: distributed team with async standups. Grade: Active.

Media and Content

Smaller category (five skills, per VoltAgent). Useful if your workflow involves publishing, thin otherwise.

transcript-cleaner Cleans raw Whisper or Deepgram transcripts: removes filler words, splits speakers, and timestamps headings. When to use: you have a recording and need a readable transcript without editing it by hand. Grade: Active.

social-post-drafter Takes a long-form article or note and produces platform-specific drafts for LinkedIn, X, and Bluesky. When to use: you publish long-form and want distribution handled. Grade: Mature.

Honest note: the Media and Content category is thinner than the others, and several skills that claim coverage here are abandoned wrappers around older video APIs. If you work heavily in media production, you will likely need to build your own skill.

Installing Skills from This List

Every skill above is installable from ClawHub using the stock CLI that ships with OpenClaw:

clawhub install <skill-slug>

For example:

clawhub install openclaw/heartbeat-scheduler
clawhub install test-runner-unified

Community skills use a single-slug form (test-runner-unified). Official SFAI Labs skills use the openclaw/ prefix. If you want to pin to a specific version:

clawhub install test-runner-unified@1.4.2

After install, run clawhub list to see what is loaded and clawhub update on a cadence that matches your appetite for change. For deeper setup context, see the installation guide.

How to Evaluate a Skill Before Installing

Curation helps, but nothing replaces looking at the source yourself. Before running clawhub install on a community skill, check five things:

1. Last commit date. Open the skill on ClawHub. If the last commit is older than 180 days and the issue tracker has unanswered questions, skip it.
2. Declared permissions. The SKILL.md frontmatter lists which tools the skill can use. If a “meeting notes” skill asks for file system write, read the body to find out why.
3. External dependencies. Skills that wrap an MCP server or make outbound API calls should declare those dependencies. Undeclared network calls are a red flag.
4. Star count trend. A skill with 5,000 stars that gained none in the last 90 days is drifting. Absolute count matters less than recent activity.
5. Source availability. If the skill source is not readable on ClawHub or a linked GitHub repo, do not install it.

For higher-stakes deployments (multi-user, customer data, compliance-regulated workflows), also review the security risks guide before granting a new skill access.

Contributing to This List

This is a living list. To suggest a skill:

1. Open an issue on the SFAI Labs blog repository with the skill slug, one-line description, and when-to-use trigger.
2. Include a link to the ClawHub page and your proposed grade (Active, Mature, Early).
3. Disclose any affiliation with the skill’s maintainer.

We review submissions quarterly and publish an updated list at the same cadence. Skills that earn inclusion stay until a review finds them stale.

Frequently Asked Questions

What makes a skill “awesome” versus just listed?

An awesome skill solves a real problem cleanly, is actively maintained, and has a clear when-to-use trigger. The ClawHub registry lists every published skill. This list cuts that down to the ones we would install on a fresh OpenClaw instance today.

What is the difference between an official OpenClaw skill and a community skill?

Official skills live in the openclaw/ namespace on ClawHub and are maintained by the OpenClaw core team or SFAI Labs engineers. Community skills are published by anyone with a GitHub account older than one week (the ClawHub publishing minimum). Treat official as the default and community as case-by-case.

Is it safe to install community skills?

Treat community skills like you would treat an npm package: the fact that it is publishable does not mean it is safe. Read the source, check the declared permissions, and apply the five-point evaluation above. The skills on this list passed that review as of the last-reviewed date at the top.

How often is this list updated?

Quarterly, with an out-of-cycle update if a listed skill goes silent or a critical skill launches. The last-reviewed date at the top of the list is authoritative.

Can I contribute a skill to this list?

Yes, see Contributing to This List. Submissions are reviewed in the next quarterly cycle.

How do I evaluate whether a skill is still maintained?

Open the skill on ClawHub, check the last commit date and the changelog, then look at open issues. If commits have stopped and issues are piling up, assume the skill is drifting toward abandonment even if it still works today.

Do these skills work with Claude Code or other agent frameworks?

Many community skills in the ecosystem are cross-compatible because the SKILL.md standard is shared, but no compatibility is guaranteed. Official SFAI Labs skills are written for OpenClaw specifically and may rely on OpenClaw-only features like the heartbeat loop.

Key Takeaways

• The ClawHub registry hosts 13,729 skills. This list shortens that to the ~25 we would install on a fresh instance today.
• Official SFAI Labs skills (the openclaw/ namespace) are the safe default. Community skills need per-install review.
• Every entry has a one-line when-to-use trigger so you can match a skill to a current situation, not just a category.
• The grading system (Active, Mature, Early) reflects maintenance signal, not popularity.
• The list is reviewed quarterly. If a skill you rely on is missing, submit it for the next cycle.
• Evaluate every community skill on five criteria before installing: last commit, permissions, dependencies, star trend, source availability.