Top agent skills for OpenClaw & Hermes
Skills are reusable capabilities you plug into an agent — and the registries now list tens of thousands of them. Most you'll never need. These are the ones worth installing first, the ones that genuinely change what the agent can do. We'll keep adding to this list.
Each page is framework-agnostic: it covers how the skill works on both OpenClaw and Hermes, because the interesting part is usually where the two frameworks differ — what's an installable skill on one is often native on the other.
Obsidian
Turn an Obsidian vault into a read-write knowledge base your agent can search, update, and build on. How the Obsidian skill works on OpenClaw vs Hermes, and how to drive it from your phone.
A read-write knowledge base for your agent
GitHub
Let your agent open pull requests, triage issues, and manage repositories. How the GitHub skill works on OpenClaw vs Hermes, and how to drive it from your phone.
Open PRs, triage issues, manage repos
Persistent Memory
Give your agent memory that survives between sessions. How persistent-memory skills work on OpenClaw vs Hermes — installable stack vs native learning loop — and how to manage them from your phone.
Native in Hermes · installable on OpenClaw
Self-Improving Agent
Let your agent write and refine its own skills from experience. How self-improvement works on OpenClaw (an installable skill) vs Hermes (built-in learning loop), and how to run it from your phone.
Native in Hermes · installable on OpenClaw
Skill Security
Skill registries have shipped malicious packages. How to vet skills before they run — skill-vetter and SkillScan on OpenClaw, threat-scanned hubs on Hermes — and how to keep approval on your phone.
Vet skills before they run with real access
Run OpenClaw & Hermes from your iPhone
Onepilot gives you an SSH terminal to deploy these agents, install skills, and approve what they run — from your pocket.