Skill

openclaw-dev-knowledgebase

Use this skill when the user asks about OpenClaw architecture overview, how OpenClaw works internally, session model, channel concepts, workspace structure, agent routing internals, plugin API design (openclaw.plugin.json, api.register*), agent configuration schema (agents.list[], bindings), multi-agent delegation model, SOUL.md/AGENTS.md/USER.md persona design, memory search internals, cron/heartbeat mechanisms, hooks/webhooks architecture, browser automation model, tool policy design, sandbox security model, node pairing protocol, discovery protocol, CLI command reference, source code structure, building from source, testing, releasing, or any OpenClaw internals/theory question. Also use for: 'create an agent', 'scaffold agent', 'create a plugin', 'scaffold plugin', 'sync knowledge', 'update knowledge base'. For hands-on operations (install, debug, configure, fix, diagnose, set up networking) use openclaw-node-operations instead.

From openclaw-dev

Install

Run in your terminal

npx claudepluginhub arctrany/openclaw-dev

Tool Access

This skill uses the workspace's default tool permissions.

Supporting Assets

View in Repository

agents/openai.yaml

references/agent-config.md

references/core-concepts.md

references/extensions-and-skills.md

references/fault-patterns.md

references/hooks-webhooks-heartbeat.md

references/install-and-debug.md

references/log-analysis-methodology.md

references/multi-node-networking.md

references/networking.md

references/plugin-api.md

references/plugin-examples.md

references/plugin-management.md

references/runbooks.md

references/scaffold-agent-guide.md

references/scaffold-plugin-guide.md

references/sessions-memory-automation-security.md

references/source-code-map.md

references/sync-knowledge-runbook.md

references/system-prompt-examples.md

Skill Content

Similar Skills

skill-lookup

Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.

prompts.chat

157.6k

prompt-lookup

Searches prompts.chat for AI prompt templates by keyword or category, retrieves by ID with variable handling, and improves prompts via AI. Use for discovering or enhancing prompts.

prompts.chat

157.6k

agent-eval

Compares coding agents like Claude Code and Aider on custom YAML-defined codebase tasks using git worktrees, measuring pass rate, cost, time, and consistency.

ecc

145.8k

Stats

Stars2

Forks1

Last CommitMar 26, 2026

Actions

View Source View Plugin View on GitHub View README

OpenClaw Dev Knowledgebase

OpenClaw 全面知识库 — 功能/架构/开发/部署/运维 + plugin API + agent 配置。

⛔ 铁律: 不可破坏 Memory

绝对不能删除、覆盖、截断 memory/ 目录下的任何文件和 MEMORY.md

只允许 append 操作，不允许 rewrite 或 truncate

迁移 workspace 时必须完整保留 memory/ 和 MEMORY.md

⛔ 铁律: 遇到问题先跑 openclaw doctor

任何异常先运行 openclaw doctor，它会自动检测并修复常见问题

⛔ 铁律: 零硬编码（组织铁律 #1）

Skill 文件内禁止写死：路径（/Users/、/Volumes/）、邮箱、IP:Port、API Key、模型名

运行时值的权威来源：~/.openclaw/openclaw.env（env vars）、~/.openclaw/openclaw.json（config）

实验室路径（/Volumes/EXT/openclaw-god/）禁止出现在生产侧 workspace 的任何文件中

知识索引

核心概念

主题	参考文件
Node / Workspace / Agent / Model	`references/core-concepts.md`
Sessions / Memory / Automation / Security	`references/sessions-memory-automation-security.md`
Hooks / Webhooks / Heartbeat	`references/hooks-webhooks-heartbeat.md`
Tools / Browser / Plugins	`references/tools-browser-plugins.md`
Workspace / Channels / Discovery	`references/workspace-channels-discovery.md`

开发指南

主题	参考文件
Plugin API (openclaw.plugin.json, api.register*)	`references/plugin-api.md`
Plugin 示例和故障排除	`references/plugin-examples.md`
Agent 配置 (agents.list[], bindings, security)	`references/agent-config.md`
System Prompt 示例 (SOUL.md, AGENTS.md, USER.md)	`references/system-prompt-examples.md`

运维参考

主题	参考文件
安装和调试 (macOS/Linux/Windows)	`references/install-and-debug.md`
多节点组网 (Tailscale, SSH, 远程 Node)	`references/multi-node-networking.md`
网络模型	`references/networking.md`
操作手册 (Runbooks)	`references/runbooks.md`

运行时分析 (活文档)

主题	参考文件
日志分析方法论 (5 步系统分析)	`references/log-analysis-methodology.md`
故障模式库 (已知模式签名, agent 可追加)	`references/fault-patterns.md`

源码参考

主题	参考文件
源码目录结构	`references/source-code-map.md`
扩展和技能	`references/extensions-and-skills.md`

操作指南 (Runbooks)

操作	参考文件
创建新 Agent (交互式 scaffold)	`references/scaffold-agent-guide.md`
创建新 Plugin (交互式 scaffold)	`references/scaffold-plugin-guide.md`
Plugin 运维管理 (安装/卸载/升级/诊断)	`references/plugin-management.md`
同步知识库 (与上游文档对齐)	`references/sync-knowledge-runbook.md`

核心架构

Gateway (控制面, 单进程)
├── Agents (多个, 各有独立 workspace/sessions)
├── Channels (WhatsApp, Telegram, Discord, iMessage...)
├── Plugins (TypeScript 扩展: tools/channels/providers)
├── Nodes (配对设备: exec/screen/canvas/camera)
└── Sessions (DM 隔离, 每次对话一个 session)

关键路径

路径	说明
`~/.openclaw/openclaw.json`	主配置
`~/.openclaw/agents/<id>/sessions/`	Session 日志
`~/.openclaw/workspace-<id>/`	Agent workspace
`~/.openclaw/extensions/`	全局 plugin 目录

常用命令

# 状态
openclaw health
openclaw status --deep --all
openclaw doctor

# Agents
openclaw agents list --bindings
jq '.agents.list[] | {id, model, workspace}' ~/.openclaw/openclaw.json

# Channels
openclaw channels status --probe

# Plugins
openclaw plugins list

# Gateway 管理
openclaw gateway install
openclaw gateway start | stop | restart

安装

平台	命令
macOS / Linux	`curl -fsSL https://openclaw.ai/install.sh \| bash`
Windows (WSL2)	`iwr -useb https://openclaw.ai/install.ps1 \| iex`
无 root	`curl -fsSL https://openclaw.ai/install-cli.sh \| bash`

Plugin 开发快速入门

# 1. 创建目录 + manifest + package.json
mkdir my-plugin && cd my-plugin
cat > openclaw.plugin.json << 'EOF'
{"id": "my-plugin", "name": "My Plugin", "description": "What it does", "configSchema": {"type": "object", "additionalProperties": false, "properties": {}}}
EOF
cat > package.json << 'EOF'
{"name": "my-plugin", "version": "1.0.0", "type": "module", "openclaw": {"extensions": ["./index.ts"]}}
EOF

# 2. TypeScript entry (必须在根目录，不能放 src/)
cat > index.ts << 'EOF'
import { Type } from "@sinclair/typebox";
import { definePluginEntry } from "openclaw/plugin-sdk/plugin-entry";

export default definePluginEntry({
  id: "my-plugin",
  name: "My Plugin",
  description: "What it does",
  register(api) {
    api.registerTool({
      name: "my_tool",
      description: "My tool",
      parameters: Type.Object({ input: Type.String() }),
      async execute(_toolCallId, params) {
        return { content: [{ type: "text", text: params.input }] };
      },
    });
  },
});
EOF

# 3. 链接安装 (开发模式)
openclaw plugins install -l .

Agent 配置快速入门

// ~/.openclaw/openclaw.json → agents.list[]
{
  id: "my-agent",
  name: "My Agent",
  workspace: "~/.openclaw/workspace-my-agent",
  model: "anthropic/claude-sonnet-4-5",
  subagents: { allowAgents: ["worker-1"] },
}

Workspace 引导文件: SOUL.md (身份) / AGENTS.md (委派) / USER.md (用户偏好)

Skill 解析顺序

Workspace skills  (最高优先)
  └── ~/.openclaw/workspace-<agent>/skills/
Managed skills    (中)
  └── ~/.openclaw/skills/ (shared)
Bundled skills    (最低)
  └── 内置于 OpenClaw