安全

Setup

Configure OpenClaw installations with optimized settings, channel setup, security hardening, and production recommendations.

它能做什么

| Task | Load | |------|------| | Messaging channels (Telegram, WhatsApp, Discord, etc.) | | | Agent settings, models, workspaces | | | Security, auth, DM policies, allowlists | | | Tools: exec, browser, web, media | | | Cron, hooks, heartbeats, automation | | | Recommendations by use case |…

技能文档

Quick Reference

TaskLoad
Messaging channels (Telegram, WhatsApp, Discord, etc.)channels.md
Agent settings, models, workspacesagents.md
Security, auth, DM policies, allowlistssecurity.md
Tools: exec, browser, web, mediatools.md
Cron, hooks, heartbeats, automationautomation.md
Recommendations by use caserecommendations.md
Memory search, embeddings, QMDmemory.md
Gateway: port, TLS, Tailscale, remotegateway.md

First Setup Checklist

Before any config, run:

openclaw onboard --install-daemon  # Full wizard
openclaw doctor                    # Check issues

Minimum viable config:

  • At least one channel connected (Telegram recommended for testing)
  • Model configured (Anthropic Claude or OpenAI)
  • Workspace path set (agents.defaults.workspace)
  • Owner allowlist configured (your user ID in channels.*.allowFrom)

Config Locations

FilePurpose
~/.openclaw/openclaw.jsonMain config
~/.openclaw/.envEnvironment variables
~/.openclaw/workspace/Default workspace
~/.openclaw/sessions/Session storage

Hot reload: Most settings apply instantly. Gateway settings (port, TLS) require restart.

Common Mistakes to Avoid

  1. Open DM policy without allowlist → Anyone can message your bot
  2. No auth token on remote gateway → Exposed to internet
  3. Model without fallbacks → Single point of failure
  4. Heartbeat without delivery target → Lost proactive messages
  5. exec.security: "full" in groups → Dangerous command access

When You're Done

openclaw doctor       # Verify config
openclaw status       # Check runtime
openclaw health       # Gateway health

相关技能

OpenClaw CLI

按命令族和 profile 选择合适的 OpenClaw CLI 命令,并校验执行结果。

334 次安装6 星标

clawsec-feed

为 AI 智能体订阅社区维护的安全公告情报源，及时获取威胁与漏洞更新。

作者 davida-ps103 次安装1 星标

Cyber Security Engineer

Manage and enforce least-privilege execution, approval-based elevation, port and egress monitoring, and ISO 27001/NIST compliance reporting for OpenClaw secu...

60 次安装

OpenClaw Dashboard

运维与扩展 OpenClaw 只读仪表盘的后端、前端标签页与安全模型。

174 次安装9 星标

openclaw-nextcloud

用一条 CLI 命令管理 Nextcloud 的笔记、任务、日历、文件、联系人和 Deck 看板。

187 次安装9 星标

Iván 的更多技能

浏览全部技能

Git

执行 Git 操作(提交、分支、合并、变基、冲突解决与恢复)时强制套用安全规则。

作者 Iván527 次安装31 星标

Design

用可量化的层级、间距、字号、配色与版式规则，绘制并诊断视觉作品。

作者 Iván137 次安装5 星标

CSS

围绕 CSS 机制排查问题并编写组件样式表,而不是凭感觉试错。

作者 Iván97 次安装5 星标

Learn

以系统方式规划并执行自学:从出口测试倒推课程,加入间隔复习与刻意练习,产出可验证的迁移证据。

作者 Iván93 次安装3 星标

Azure

针对你的 Azure 订阅,做架构设计、故障排查、安全加固与成本优化

作者 Iván86 次安装2 星标

Java

按配置的 JDK 版本诊断 Java 与 JVM 问题(从 NPE 到容器 OOM),给出可直接套用的代码与配置。

作者 Iván130 次安装9 星标