菜单
Builds OpenClaw plugins for NemoClaw — native plugins with manifests and runtime modules, or bundle plugins in Codex/Claude/Cursor format. Use when extending NemoClaw agents with new providers, tools, channels, or skills.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Reviews NemoClaw sandbox security policies for gaps, overly permissive rules, and best practice violations. Use when auditing policy security, before deploying policies to production, or during security reviews.
Generates and applies OpenShell YAML security policies for NemoClaw sandboxes — filesystem, network, process, and inference controls. Use when creating, editing, applying, or reviewing sandbox security policies, network egress rules, or filesystem permissions.
Scaffolds new OpenClaw AgentSkills for use inside NemoClaw sandboxes, with proper SKILL.md frontmatter, directory structure, and conditional gating. Use when building custom skills for NemoClaw agents or when the user mentions creating, authoring, or scaffolding agent skills.
Diagnoses NemoClaw agent issues across sandbox, policy, and inference layers. Use when agents behave unexpectedly, sandbox won't start, inference fails, network requests are blocked, or plugins report errors.
Discovers, installs, and publishes OpenClaw skills via ClawHub for use in NemoClaw sandboxes. Use when finding community skills, installing from ClawHub, publishing custom skills, or browsing the skills marketplace.
Manages credential providers for NemoClaw sandboxes — create, inspect, update, and delete API key providers for NVIDIA, Claude, GitHub, OpenAI, Ollama, and generic APIs. Use when setting up inference credentials, API access, or rotating keys for NemoClaw.
| name | creating-nemoclaw-plugins |
| description | Builds OpenClaw plugins for NemoClaw — native plugins with manifests and runtime modules, or bundle plugins in Codex/Claude/Cursor format. Use when extending NemoClaw agents with new providers, tools, channels, or skills. |
OpenClaw plugins extend agent capabilities by bundling channels, model providers, tools, and skills into installable packages. Plugins come in two formats.
Creating new capabilities from scratch? Use a native plugin. Adapting an existing Codex/Claude/Cursor plugin? Use a bundle plugin.
Native plugins use a openclaw.plugin.json manifest and a runtime module that registers capabilities.
my-plugin/
├── openclaw.plugin.json # Manifest
├── index.ts # Runtime entry point
├── package.json # npm package config
└── skills/ # Optional bundled skills
Use the bundled template: native-manifest.json
import { definePluginEntry } from "@openclaw/sdk";
export default definePluginEntry({
id: "my-plugin",
register(api) {
api.registerProvider({ /* model provider config */ });
api.registerTool({ /* tool definition */ });
api.registerChannel({ /* messaging channel */ });
}
});
See plugin-api.md for the full registration API.
Bundle plugins use directory layouts compatible with Codex, Claude, or Cursor:
.codex-plugin/ # Codex-compatible
.claude-plugin/ # Claude-compatible
.cursor-plugin/ # Cursor-compatible
See bundle-structure.md for layout details.
All plugin management commands run inside the sandbox:
# Connect to sandbox
nemoclaw <name> connect
# Link plugin for development (no copy, changes reflect immediately)
openclaw plugins install -l /path/to/my-plugin
# Check plugin health
openclaw plugins doctor
# Inspect plugin details
openclaw plugins inspect my-plugin
# Enable/disable
openclaw plugins enable my-plugin
openclaw plugins disable my-plugin
Plugins installed inside the sandbox are lost on restart unless:
read_only or read_write path)