OpenClaw完整命令列表(更新至2026.4.21版本)
OpenClaw Available Commands
Tools
/help — Show available commands.
/commands — List all slash commands. (agent)
/tools [mode] — List available runtime tools. (agent)
/skill [input] — Run a skill by name. (agent)
/status — Show current status. (agent)
/tasks — List background tasks for this session. (agent)
/allowlist — List/add/remove allowlist entries. (agent)
/approve — Approve or deny exec requests. (agent)
/context — Explain how context is built and used. (agent)
/btw — Ask a side question without changing future session context. (agent)
/export-session [path] — Export current session to HTML file with full system prompt.
/tts [action] [value] — Control text-to-speech (TTS). (agent)
/whoami — Show your sender id. (agent)
/acp [action] [value] — Manage ACP sessions and runtime options. (agent)
/config [action] [path] [value] — Show or set config values. (agent)
/mcp [action] [path] [value] — Show or set OpenClaw MCP servers. (agent)
/plugins [action] [path] — List, show, enable, or disable plugins. (agent)
/debug [action] [path] [value] — Set runtime debug overrides. (agent)
/usage [mode] — Usage footer or cost summary.
/restart — Restart OpenClaw. (agent)
/activation [mode] — Set group activation mode. (agent)
/send [mode] — Set send policy. (agent)
/bash [command] — Run host shell commands (host-only). (agent)
Session
/session [action] [value] — Manage session-level settings (for example /session idle). (agent)
/focus [target] — Bind this thread (Discord) or topic/conversation (Telegram) to a session target.
/unfocus — Remove the current thread (Discord) or topic/conversation (Telegram) binding. (agent)
/stop — Stop the current run.
/reset — Reset the current session.
/new — Start a new session.
/compact [instructions] — Compact the session context.
/clear — Clear chat history
Agents
/subagents [action] [target] [value] — List, kill, log, spawn, or steer subagent runs for this session. (agent)
/agents — List thread-bound agents for this session.
/kill [target] — Kill a running subagent (or all).
/steer [id] — Inject a message into the active run
/redirect [id] — Abort and restart with a new message
Model
/think [level] — Set thinking level.
/verbose [mode] — Toggle verbose mode.
/trace [mode] — Toggle plugin trace lines. (agent)
/fast [mode] — Toggle fast mode.
/reasoning [mode] — Toggle reasoning visibility. (agent)
/elevated [mode] — Toggle elevated mode. (agent)
/exec [host] [security] [ask] [node] — Set exec defaults for this session. (agent)
/model [model] — Show or set the model.
/models — List model providers or provider models. (agent)
/queue [mode] [debounce] [cap] [drop] — Adjust queue settings. (agent)
Type / to open the command menu.
OpenClaw 的完整命令列表
这是 OpenClaw 的完整命令列表翻译。为了方便查阅,我按照官方文档的分类进行了整理,并保留了命令的原始格式。
工具 (Tools)
- /help — 显示可用命令列表。
- /commands — 列出所有斜杠命令。(代理端)
- /tools [模式] — 列出可用的运行时工具。(代理端)
- /skill [输入] — 按名称运行某个技能。(代理端)
- /status — 显示当前状态。(代理端)
- /tasks — 列出此会话的后台任务。(代理端)
- /allowlist — 列出/添加/移除白名单条目。(代理端)
- /approve — 批准或拒绝执行请求。(代理端)
- /context — 解释上下文是如何构建和使用的。(代理端)
- /btw — 问一个侧面问题,不改变未来会话的上下文。(代理端)
- /export-session [路径] — 将会话导出为包含完整系统提示词的 HTML 文件。
- /tts [动作] [值] — 控制文本转语音。
- /whoami — 显示你的发送者 ID。
- /acp [动作] [值] — 管理 ACP 会话和运行时选项。(代理端)
- /config [动作] [路径] [值] — 显示或设置配置值。(代理端)
- /mcp [动作] [路径] [值] — 显示或设置 OpenClaw MCP 服务器。(代理端)
- /plugins [动作] [路径] — 列出、显示、启用或禁用插件。(代理端)
- /debug [动作] [路径] [值] — 设置运行时调试覆盖。
- /usage [模式] — 显示使用量页脚或费用摘要。
- /restart — 重启 OpenClaw。(代理端)
- /activation [模式] — 设置群组激活模式。(代理端)
- /send [模式] — 设置发送策略。(代理端)
- /bash [命令] — 运行宿主机 Shell 命令(仅限宿主机)。
会话 (Session)
- /session [动作] [值] — 管理会话级设置(例如
/session idle)。 - /focus [目标] — 将此线程或话题绑定到会话目标。
- /unfocus — 移除当前线程或话题的绑定。
- /stop — 停止当前运行。
- /reset — 重置当前会话。
- /new — 开始新会话。
- /compact [指令] — 压缩会话上下文。
- /clear — 清除聊天历史。
代理 (Agents)
- /subagents [动作] [目标] [值] — 列出、终止、记录日志、生成或引导此会话的子代理运行。
- /agents — 列出此会话中绑定到线程的代理。
- /kill [目标] — 终止正在运行的子代理(或全部)。
- /steer [ID] — 向正在运行的代理注入一条消息。
- /redirect [ID] — 中止当前运行并用新消息重新启动。
模型 (Model)
- /think [等级] — 设置思考等级。
- /verbose [模式] — 切换详细模式。
- /trace [模式] — 切换插件追踪行显示。(代理端)
- /fast [模式] — 切换快速模式。
- /reasoning [模式] — 切换推理过程可见性。(代理端)
- /elevated [模式] — 切换提升模式。(代理端)
- /exec [宿主机] [安全] [询问] [节点] — 设置此会话的执行默认值。
- /model [模型] — 显示或设置模型。
- /models — 列出模型提供商或提供商模型。
- /queue [模式] [防抖] [上限] [丢弃] — 调整队列设置。
提示: 在输入框中输入 / 即可打开命令菜单。
AtomGit 是由开放原子开源基金会联合 CSDN 等生态伙伴共同推出的新一代开源与人工智能协作平台。平台坚持“开放、中立、公益”的理念,把代码托管、模型共享、数据集托管、智能体开发体验和算力服务整合在一起,为开发者提供从开发、训练到部署的一站式体验。
更多推荐

所有评论(0)