Live survey286 models · 13 providers · 17,592 auditsAudited Using Git Worktrees · SAFE →What changedHow we verifyllms.txt

Agent host

OpenClaw

967 skills · 0 MCP serversLatest audit 28 Oct 2026

Skills · 967
967 audited agent skills on ModelsAtlas mention OpenClaw: 76% are stamped SAFE and 140 are BLOCK and should not be installed. No MCP server record carries a OpenClaw connect snippet yet. Below: how to add each kind, and the safest picks.
Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
More
CoCopilotAIGoogle AI Mode
Cite

Cite this page

ModelsAtlas. “OpenClaw: skills and MCP servers”. Retrieved 28 Oct 2026. https://modelsatlas.com/hosts/openclaw
Skills
967

name OpenClaw in their docs

SAFE skills
734

76% · 93 caution · 140 block

MCP servers
0

no snippet yet

MCP BLOCK
0

none listed

Works with

OpenClaw in the atlas

▣Agent skills

Install a skill in OpenClaw

Only what each skill's own page records

  1. Check the stamp first

    Open the skill on ModelsAtlas. Install SAFE skills freely, read the findings on CAUTION, never install BLOCK.

  2. Use the audited commit

    Every audit names the commit it read. The stamp covers that commit only, not the branch head.

  3. Run what the skill says

    Where a skill's instructions give an install line, its page lists it. Where they do not, we show nothing rather than invent one.

npx skills add https://github.com/mem0ai/mem0 --skill mem0-vercel-ai-sdk

From the instructions of Mem0 Vercel AI SDK Skill for Claude (mem0ai/mem0), the most-starred SAFE skill here that gives a complete install line. Audited commit 5501ce5.

Connect an MCP server

Snippets come from each server's published package

No MCP server record carries a connect snippet for OpenClaw yet, so we show no command. Servers with a snippet for other clients are on the MCP servers hub.

Safest skills for OpenClaw

All 967 OpenClaw skills →

SAFE skills only, highest audit score first, then stars.

Social Publisher

affaan-m/ECC

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

✓SAFE B★ 264,831

Openclaw Persona Forge

affaan-m/ECC

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

✓SAFE B★ 263,919

Honcho

NousResearch/hermes-agent

The agent that grows with you

✓SAFE B★ 247,828

Hyperframes

NousResearch/hermes-agent

The agent that grows with you

✓SAFE B★ 247,828

Browser Use

browser-use/browser-use

Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.

✓SAFE B★ 118,614

Create Video

calesthio/openmontage

World's first open-source, agentic video production system. 12 production pipelines, 100+ tools, 700+ agent skill and production-knowledge files. Turn your AI coding assistant into a full video production studio.

✓SAFE B★ 72,344

Questions about OpenClaw

How do I install a skill in OpenClaw?
Open the skill's page on ModelsAtlas. It shows the install lines the skill's own instructions give, when it has any, and the commit we audited. Install SAFE skills as you would any dependency, read the findings on CAUTION first, and do not install BLOCK. The stamp covers the audited commit only, so use that commit rather than the branch head.
Are OpenClaw skills safe?
Most are not a problem, but check each one: of 967 audited skills that mention OpenClaw, 734 are SAFE, 93 CAUTION and 140 BLOCK. A skill can run scripts and steer the agent, so look at its stamp before you install.
Which skills work in OpenClaw?
The 967 that name OpenClaw in their own documentation. That is what the text claims, not a compatibility test we ran.
Can I connect MCP servers to OpenClaw?
We have no MCP server record with a connect snippet for OpenClaw yet, so we show none rather than guess a command.

Keep going in your AI

Opens with this page as the source. We add nothing else to the prompt.