content large_stringlengths 3 20.5k | url large_stringlengths 54 193 | branch large_stringclasses 4
values | source large_stringclasses 42
values | embeddings listlengths 384 384 | score float64 -0.21 0.65 |
|---|---|---|---|---|---|
# 安全性 🔒 ## 快速检查：`openclaw security audit` 另请参阅：[形式化验证（安全模型）](/security/formal-verification/) 定期运行此命令（尤其是在更改配置或暴露网络接口之后）： ```bash openclaw security audit openclaw security audit --deep openclaw security audit --fix ``` 它会标记常见的安全隐患（Gateway 网关认证暴露、浏览器控制暴露、提权白名单、文件系统权限）。 `--fix` 会应用安全防护措施： - 将常见渠道的 `groupPolicy="open"` 收紧... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/gateway/security/index.md | main | opebclaw | [
-0.07250703871250153,
0.06789474189281464,
-0.08893617987632751,
0.026262348517775536,
0.045765623450279236,
-0.058173686265945435,
0.08801393955945969,
-0.060753922909498215,
-0.0023053118493407965,
-0.0170438289642334,
0.11289293318986893,
-0.04751593619585037,
0.05721520632505417,
-0.02... | 0.098328 |
假设"被入侵"意味着：有人进入了可以触发机器人的房间，或者令牌泄露，或者插件/工具做了意外的事情。 1. **阻止影响范围** - 禁用提权工具（或停止 Gateway 网关）直到你了解发生了什么。 - 锁定入站接口（私信策略、群组白名单、提及门控）。 2. **轮换秘密** - 轮换 `gateway.auth` 令牌/密码。 - 轮换 `hooks.token`（如果使用）并撤销任何可疑的节点配对。 - 撤销/轮换模型提供商凭证（API 密钥/OAuth）。 3. **审查产物** - 检查 Gateway 网关日志和最近的会话/记录中是否有意外的工具调用。 - 审查 `extensions/` 并移除任何你不完全信任的内容。 ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/gateway/security/index.md | main | opebclaw | [
-0.041089847683906555,
-0.0346553809940815,
-0.038207925856113434,
-0.05849875509738922,
-0.0015626333188265562,
-0.06180773675441742,
0.06637808680534363,
0.00427794735878706,
0.03825758025050163,
-0.04198368638753891,
0.11405923962593079,
-0.01239711046218872,
0.04842362552881241,
-0.005... | 0.035269 |
"loopback", port: 18789, auth: { mode: "token", token: "your-long-random-token" }, }, channels: { whatsapp: { dmPolicy: "pairing", groups: { "*": { requireMention: true } }, }, }, } ``` 如果你还想要"默认更安全"的工具执行，为任何非所有者智能体添加沙箱 + 拒绝危险工具（示例见下方"单智能体访问配置"）。 ## 沙箱隔离（推荐） 专用文档：[沙箱隔离](/gateway/sandboxing) 两种互补的方法： - **在 Docker 中运行完整的... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/gateway/security/index.md | main | opebclaw | [
-0.045652393251657486,
0.05152348801493645,
-0.03622790798544884,
-0.09902338683605194,
0.029378876090049744,
-0.06877409666776657,
0.05037384480237961,
-0.04655349999666214,
0.006452356930822134,
0.02783021703362465,
0.09912187606096268,
-0.021634100005030632,
-0.011235281825065613,
0.026... | 0.074493 |
# Node + tsx "\\_\\_name is not a function" 崩溃 ## 概述 通过 Node 使用 `tsx` 运行 OpenClaw 时，启动阶段报错： ``` [openclaw] Failed to start CLI: TypeError: \_\_name is not a function at createSubsystemLogger (.../src/logging/subsystem.ts:203:25) at .../src/agents/auth-profiles/constants.ts:25:20 ``` 此问题在开发脚本从 Bun 切换到 `tsx` 后出现（提交 `2871... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/debug/node-issue.md | main | opebclaw | [
-0.051028773188591,
0.017433950677514076,
-0.04369571805000305,
-0.061301037669181824,
0.0342523530125618,
-0.07911886274814606,
0.0053030760027468204,
0.005975967273116112,
-0.00820742268115282,
0.03805944696068764,
0.04817364364862442,
-0.09446576237678528,
0.07399722188711166,
-0.015671... | 0.091149 |
# BlueBubbles（macOS REST） 状态：内置插件，通过 HTTP 与 BlueBubbles macOS 服务器通信。由于其更丰富的 API 和更简便的设置，\*\*推荐用于 iMessage 集成\*\*，优于旧版 imsg 渠道。 ## 概述 - 通过 BlueBubbles 辅助应用在 macOS 上运行（[bluebubbles.app](https://bluebubbles.app)）。 - 推荐/已测试版本：macOS Sequoia (15)。macOS Tahoe (26) 可用；但在 Tahoe 上编辑功能目前不可用，群组图标更新可能显示成功但实际未同步。 - OpenClaw 通过其 REST... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/bluebubbles.md | main | opebclaw | [
-0.039405759423971176,
0.001680346904322505,
0.04933403432369232,
-0.03995893523097038,
-0.0017881228122860193,
-0.09850800782442093,
0.05067029595375061,
-0.025586450472474098,
0.04334264621138573,
-0.018774565309286118,
0.09757863730192184,
-0.028671540319919586,
0.0819450169801712,
-0.0... | 0.181157 |
相关全局选项： - `agents.list[].groupChat.mentionPatterns`（或 `messages.groupChat.mentionPatterns`）。 - `messages.responsePrefix`。 ## 地址 / 投递目标 优先使用 `chat_guid` 以获得稳定的路由： - `chat_guid:iMessage;-;+15555550123`（群组推荐） - `chat_id:123` - `chat_identifier:...` - 直接句柄：`+15555550123`、`user@example.com` - 如果直接句柄没有现有的私信聊天，OpenClaw 将通过 `P... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/bluebubbles.md | main | opebclaw | [
-0.045949652791023254,
0.0032255216501653194,
0.006741462741047144,
-0.0017162543954327703,
-0.024160223081707954,
-0.05352276191115379,
0.08232919871807098,
-0.0448102168738842,
0.06232185661792755,
-0.03283820301294327,
0.08543004095554352,
-0.06989429146051407,
0.07137526571750641,
0.00... | 0.068444 |
# 飞书机器人 状态：生产就绪，支持机器人私聊和群组。使用 WebSocket 长连接模式接收消息。 --- ## 需要插件 安装 Feishu 插件： ```bash openclaw plugins install @openclaw/feishu ``` 本地 checkout（在 git 仓库内运行）： ```bash openclaw plugins install ./extensions/feishu ``` --- ## 快速开始 添加飞书渠道有两种方式： ### 方式一：通过安装向导添加（推荐） 如果您刚安装完 OpenClaw，可以直接运行向导，根据提示添加飞书： ```bash openclaw onboard... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/feishu.md | main | opebclaw | [
-0.04436681792140007,
0.04102827608585358,
-0.09079213440418243,
-0.023638896644115448,
0.008880586363375187,
-0.09391403943300247,
0.01632884331047535,
0.028664622455835342,
-0.011912223882973194,
-0.03360189124941826,
0.09820586442947388,
-0.054905541241168976,
0.029205910861492157,
-0.0... | 0.094923 |
| `/status` | 查看机器人状态 | | `/reset` | 重置对话会话 | | `/model` | 查看/切换模型 | > 注意：飞书目前不支持原生命令菜单，命令需要以文本形式发送。 ## 网关管理命令 在配置和使用飞书渠道时，您可能需要使用以下网关管理命令： | 命令 | 说明 | | -------------------------- | ----------------- | | `openclaw gateway status` | 查看网关运行状态 | | `openclaw gateway install` | 安装/启动网关服务 | | `openclaw gateway stop` | 停止网关服... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/feishu.md | main | opebclaw | [
-0.030967550352215767,
0.01341088954359293,
-0.04352310672402382,
-0.01905634067952633,
-0.021076932549476624,
-0.01886938326060772,
-0.01826850138604641,
-0.015016193501651287,
-0.0065931277349591255,
0.004339540842920542,
0.14677724242210388,
-0.020683016628026962,
0.01562405377626419,
-... | 0.090478 |
# 渠道位置解析 OpenClaw 将聊天渠道中分享的位置标准化为： - 附加到入站消息体的可读文本，以及 - 自动回复上下文负载中的结构化字段。 目前支持： - \*\*Telegram\*\*（位置图钉 + 地点 + 实时位置） - \*\*WhatsApp\*\*（locationMessage + liveLocationMessage） - \*\*Matrix\*\*（`m.location` 配合 `geo\_uri`） ## 文本格式 位置以友好的行格式呈现，不带括号： - 图钉： - `📍 48.858844, 2.294351 ±12m` - 命名地点： - `📍 Eiffel Tower — Champ d... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/location.md | main | opebclaw | [
0.019659554585814476,
0.06695176661014557,
0.014109160751104355,
-0.06742361932992935,
-0.0148020563647151,
-0.07536087930202484,
-0.0037791605573147535,
-0.02897731401026249,
0.01166652049869299,
0.029143910855054855,
0.12504048645496368,
-0.07864688336849213,
0.018786083906888962,
-0.045... | 0.042567 |
# 渠道故障排除 首先运行： ```bash openclaw doctor openclaw channels status --probe ``` `channels status --probe` 会在检测到常见渠道配置错误时输出警告，并包含小型实时检查（凭据、部分权限/成员资格）。 ## 渠道 - Discord：[/channels/discord#troubleshooting](/channels/discord#troubleshooting) - Telegram：[/channels/telegram#troubleshooting](/channels/telegram#troubleshooting) - W... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/troubleshooting.md | main | opebclaw | [
-0.0025448924861848354,
0.002251589437946677,
0.01536292489618063,
-0.041740819811820984,
0.028474202379584312,
-0.12053592503070831,
-0.022225167602300644,
-0.033630676567554474,
0.045558445155620575,
0.05155323073267937,
0.03744174540042877,
-0.04668305814266205,
-0.011196229606866837,
0... | 0.0575 |
# Slack ## Socket 模式（默认） ### 快速设置（新手） 1. 创建一个 Slack 应用并启用 \*\*Socket Mode\*\*。 2. 创建一个 \*\*App Token\*\*（`xapp-...`）和 \*\*Bot Token\*\*（`xoxb-...`）。 3. 为 OpenClaw 设置令牌并启动 Gateway 网关。 最小配置： ```json5 { channels: { slack: { enabled: true, appToken: "xapp-...", botToken: "xoxb-...", }, }, } ``` ### 设置 1. 在 https://api.slac... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/slack.md | main | opebclaw | [
-0.0848269909620285,
-0.04366592690348625,
0.010573399253189564,
0.026117373257875443,
-0.007590060587972403,
-0.06078151613473892,
0.03607870265841484,
0.009940803982317448,
-0.0032250406220555305,
-0.0036763933021575212,
0.041314657777547836,
-0.059869587421417236,
0.016057681292295456,
... | 0.160753 |
https://docs.slack.dev/reference/methods/reactions.add - `pins:read`、`pins:write`（`pins.list` / `pins.add` / `pins.remove`） https://docs.slack.dev/reference/scopes/pins.read https://docs.slack.dev/reference/scopes/pins.write - `emoji:read`（`emoji.list`） https://docs.slack.dev/reference/scopes/emoji.read - `files:write`... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/slack.md | main | opebclaw | [
-0.05652627348899841,
-0.05583372712135315,
0.014687093906104565,
0.007318134885281324,
0.019251881167292595,
-0.026838799938559532,
0.015778031200170517,
0.03685644641518593,
0.05122759938240051,
0.06828396022319794,
0.06920153647661209,
0.012385797686874866,
0.018527794629335403,
0.04017... | 0.145154 |
| | emojiList | 已启用 | 自定义表情符号列表 | ## 安全说明 - 写入默认使用 bot 令牌，因此状态更改操作保持在应用的机器人权限和身份范围内。 - 设置 `userTokenReadOnly: false` 允许在 bot 令牌不可用时使用用户令牌进行写入操作，这意味着操作以安装用户的访问权限运行。将用户令牌视为高权限，并保持操作门控和白名单严格。 - 如果你启用用户令牌写入，请确保用户令牌包含你期望的写入权限范围（`chat:write`、`reactions:write`、`pins:write`、`files:write`），否则这些操作将失败。 ## 说明 - 提及门控通过 `channels.sl... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/slack.md | main | opebclaw | [
-0.035330962389707565,
-0.03132251277565956,
-0.02466709353029728,
-0.024328839033842087,
-0.023959795013070107,
-0.0028927563689649105,
0.08434543013572693,
0.03950017690658569,
0.042256761342287064,
0.0060577429831027985,
0.12933272123336792,
-0.0006704034167341888,
0.11271428316831589,
... | 0.173219 |
# Twitch（插件） 通过 IRC 连接支持 Twitch 聊天。OpenClaw 以 Twitch 用户（机器人账户）身份连接，在频道中接收和发送消息。 ## 需要插件 Twitch 作为插件发布，未与核心安装捆绑。 通过 CLI 安装（npm 注册表）： ```bash openclaw plugins install @openclaw/twitch ``` 本地检出（从 git 仓库运行时）： ```bash openclaw plugins install ./extensions/twitch ``` 详情：[插件](/plugin) ## 快速设置（新手） 1. 为机器人创建一个专用的 Twitch 账户（或使用现... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/twitch.md | main | opebclaw | [
-0.09704367816448212,
-0.025580156594514847,
-0.021821938455104828,
-0.01725666970014572,
0.022968662902712822,
-0.05126481503248215,
0.04933067038655281,
0.010648517869412899,
0.0495770163834095,
-0.006761338096112013,
0.05087877810001373,
-0.08097923547029495,
0.017877012491226196,
-0.05... | 0.133742 |
`obtainmentTimestamp` - 令牌获取时间戳 - `allowFrom` - 用户 ID 允许列表 - `allowedRoles` - 基于角色的访问控制（`"moderator" | "owner" | "vip" | "subscriber" | "all"`） - `requireMention` - 需要 @提及（默认：`true`） \*\*提供商选项：\*\* - `channels.twitch.enabled` - 启用/禁用渠道启动 - `channels.twitch.username` - 机器人用户名（简化的单账户配置） - `channels.twitch.accessToken` - ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/twitch.md | main | opebclaw | [
-0.04645802453160286,
-0.02028460055589676,
-0.02901834063231945,
-0.06647205352783203,
0.008340656757354736,
-0.022716624662280083,
0.0753515288233757,
-0.02303769625723362,
0.0008351922151632607,
0.021856488659977913,
0.048619382083415985,
-0.09176202118396759,
0.04862736538052559,
0.002... | -0.022994 |
# Tlon（插件） Tlon 是一个基于 Urbit 构建的去中心化即时通讯工具。OpenClaw 连接到你的 Urbit ship，可以响应私信和群聊消息。群组回复默认需要 @ 提及，并可通过允许列表进一步限制。 状态：通过插件支持。支持私信、群组提及、话题回复和纯文本媒体回退（URL 附加到说明文字）。不支持表情回应、投票和原生媒体上传。 ## 需要插件 Tlon 作为插件提供，不包含在核心安装中。 通过 CLI 安装（npm 仓库）： ```bash openclaw plugins install @openclaw/tlon ``` 本地检出（从 git 仓库运行时）： ```bash openclaw plugins ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/tlon.md | main | opebclaw | [
-0.0423029400408268,
0.08446084707975388,
-0.052840858697891235,
-0.029328731819987297,
-0.016769107431173325,
-0.0940188616514206,
0.029691290110349655,
0.004989404696971178,
0.008384671062231064,
-0.04135248810052872,
0.10408788919448853,
-0.03788933530449867,
0.04631296917796135,
-0.037... | 0.121225 |
# Microsoft Teams（插件） > "进入此地者，放弃一切希望。" 更新时间：2026-01-21 状态：支持文本 + 私信附件；频道/群组文件发送需要 `sharePointSiteId` + Graph 权限（参见[在群聊中发送文件](#sending-files-in-group-chats)）。投票通过 Adaptive Cards 发送。 ## 需要插件 Microsoft Teams 作为插件提供，不包含在核心安装中。 \*\*破坏性变更（2026.1.15）：\*\* MS Teams 已从核心移出。如果你使用它，必须安装插件。 原因说明：保持核心安装更轻量，并让 MS Teams 依赖项可以独立更新。 通... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/msteams.md | main | opebclaw | [
-0.050131332129240036,
0.04472716525197029,
-0.05564630404114723,
-0.015736576169729233,
0.021417126059532166,
-0.034098390489816666,
-0.010060913860797882,
-0.015537002123892307,
0.025119798257946968,
0.025114312767982483,
0.1120123341679573,
-0.03075353614985943,
0.06089352071285248,
0.0... | 0.144533 |
a bot ID manually\*\* 并粘贴你的 Azure Bot App ID 5. 勾选范围：\*\*Personal\*\*、\*\*Team\*\*、\*\*Group Chat\*\* 6. 点击 \*\*Distribute\*\* → \*\*Download app package\*\* 7. 在 Teams 中：\*\*Apps\*\* → \*\*Manage your apps\*\* → \*\*Upload a custom app\*\* → 选择 ZIP 这通常比手动编辑 JSON 清单更容易。 ## 测试机器人 \*\*选项 A：Azure Web Chat（先验证 webhook）\*\*... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/msteams.md | main | opebclaw | [
-0.0954347774386406,
0.023191820830106735,
-0.028220780193805695,
-0.04973871633410454,
-0.0007228410104289651,
0.0054396577179431915,
0.0834401473402977,
-0.05791417509317398,
0.03723224252462387,
0.10407082736492157,
0.12399381399154663,
-0.11042023450136185,
0.10430832207202911,
0.02625... | 0.109103 |
否 | 是（可查询历史） | | \*\*设置复杂度\*\* | 仅应用清单 | 需要管理员同意 + 令牌流程 | | \*\*离线工作\*\* | 否（必须运行） | 是（随时查询） | \*\*结论：\*\* RSC 用于实时监听；Graph API 用于历史访问。要在离线时补上错过的消息，你需要带有 `ChannelMessage.Read.All` 的 Graph API（需要管理员同意）。 ## 启用 Graph 的媒体 + 历史（频道所需） 如果你需要\*\*频道\*\*中的图片/文件或想要获取\*\*消息历史\*\*，你必须启用 Microsoft Graph 权限并授予管理员同意。 1. 在 Entra ID（Azu... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/msteams.md | main | opebclaw | [
-0.08712512999773026,
0.017238371074199677,
-0.04572270065546036,
-0.08455313742160797,
-0.0026887734420597553,
-0.03460129350423813,
0.02488337643444538,
0.028494827449321747,
0.009201574139297009,
0.011291978880763054,
0.13503624498844147,
-0.043886203318834305,
0.08912128210067749,
0.06... | 0.100697 |
Base64 编码内联（无需 SharePoint 即可工作） | ### 文件存储位置 上传的文件存储在配置的 SharePoint 站点默认文档库中的 `/OpenClawShared/` 文件夹中。 ## 投票（Adaptive Cards） OpenClaw 将 Teams 投票作为 Adaptive Cards 发送（没有原生 Teams 投票 API）。 - CLI：`openclaw message poll --channel msteams --target conversation: ...` - 投票由 Gateway 网关记录在 `~/.openclaw/msteams-polls.json` 中。 - Ga... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/msteams.md | main | opebclaw | [
-0.0289497971534729,
0.0007966517005115747,
-0.08891195803880692,
-0.020514866337180138,
-0.005637232214212418,
-0.07204386591911316,
-0.029861995950341225,
-0.006404770538210869,
0.04673897102475166,
-0.0839235708117485,
0.07700887322425842,
-0.021561242640018463,
0.044701457023620605,
-0... | 0.124703 |
# Signal (signal-cli) 状态：外部 CLI 集成。Gateway 网关通过 HTTP JSON-RPC + SSE 与 `signal-cli` 通信。 ## 快速设置（初学者） 1. 为 bot 使用\*\*单独的 Signal 号码\*\*（推荐）。 2. 安装 `signal-cli`（需要 Java）。 3. 链接 bot 设备并启动守护进程： - `signal-cli link -n "OpenClaw"` 4. 配置 OpenClaw 并启动 Gateway 网关。 最小配置： ```json5 { channels: { signal: { enabled: true, account: "+15... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/signal.md | main | opebclaw | [
-0.11037386208772659,
-0.01949666067957878,
0.003096525091677904,
-0.08151112496852875,
0.015087691135704517,
-0.08527430891990662,
-0.04073959216475487,
-0.021696222946047783,
0.010005266405642033,
-0.058925215154886246,
0.08096156269311905,
-0.03401349484920502,
-0.011990214698016644,
0.... | 0.082114 |
# Zalo (Bot API) 状态：实验性。仅支持私信；根据 Zalo 文档，群组即将推出。 ## 需要插件 Zalo 以插件形式提供，不包含在核心安装中。 - 通过 CLI 安装：`openclaw plugins install @openclaw/zalo` - 或在新手引导期间选择 \*\*Zalo\*\* 并确认安装提示 - 详情：[插件](/plugin) ## 快速设置（初学者） 1. 安装 Zalo 插件： - 从源代码检出：`openclaw plugins install ./extensions/zalo` - 从 npm（如果已发布）：`openclaw plugins install @openclaw... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/zalo.md | main | opebclaw | [
-0.11570243537425995,
0.017875663936138153,
-0.09093905240297318,
0.009863026440143585,
-0.019972488284111023,
-0.09872768074274063,
0.015276182442903519,
-0.002517060376703739,
-0.03563753515481949,
-0.0007809576345607638,
0.1110399067401886,
-0.07150938361883163,
0.0959639698266983,
0.01... | 0.189477 |
# grammY 集成（Telegram Bot API） # 为什么选择 grammY - 以 TS 为核心的 Bot API 客户端，内置长轮询 + webhook 辅助工具、中间件、错误处理和速率限制器。 - 媒体处理辅助工具比手动编写 fetch + FormData 更简洁；支持所有 Bot API 方法。 - 可扩展：通过自定义 fetch 支持代理，可选的会话中间件，类型安全的上下文。 # 我们发布的内容 - \*\*单一客户端路径：\*\* 移除了基于 fetch 的实现；grammY 现在是唯一的 Telegram 客户端（发送 + Gateway 网关），默认启用 grammY throttler。 - \*\*... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/grammy.md | main | opebclaw | [
-0.09035661816596985,
0.03555402532219887,
-0.024450689554214478,
-0.03894814848899841,
-0.018809404224157333,
-0.052978526800870895,
0.09255415201187134,
0.014432072639465332,
0.03875043988227844,
0.019191978499293327,
0.08819078654050827,
-0.08738811314105988,
0.08522149175405502,
-0.038... | 0.159856 |
# 聊天渠道 OpenClaw 可以在你已经使用的任何聊天应用上与你交流。每个渠道通过 Gateway 网关连接。 所有渠道都支持文本；媒体和表情回应的支持因渠道而异。 ## 支持的渠道 - [WhatsApp](/channels/whatsapp) — 最受欢迎；使用 Baileys，需要二维码配对。 - [Telegram](/channels/telegram) — 通过 grammY 使用 Bot API；支持群组。 - [Discord](/channels/discord) — Discord Bot API + Gateway；支持服务器、频道和私信。 - [Slack](/channels/slack) — Bol... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/index.md | main | opebclaw | [
-0.0681602954864502,
-0.054195236414670944,
0.042261119931936264,
-0.06496448814868927,
0.049793027341365814,
-0.08338630944490433,
-0.003157661994919181,
-0.021854449063539505,
0.06389510631561279,
-0.009543525986373425,
0.07102292776107788,
-0.058712854981422424,
0.015550992451608181,
-0... | 0.092622 |
# WhatsApp（网页渠道） 状态：仅支持通过 Baileys 的 WhatsApp Web。Gateway 网关拥有会话。 ## 快速设置（新手） 1. 如果可能，使用\*\*单独的手机号码\*\*（推荐）。 2. 在 `~/.openclaw/openclaw.json` 中配置 WhatsApp。 3. 运行 `openclaw channels login` 扫描二维码（关联设备）。 4. 启动 Gateway 网关。 最小配置： ```json5 { channels: { whatsapp: { dmPolicy: "allowlist", allowFrom: ["+15551234567"], }, }, } `... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/whatsapp.md | main | opebclaw | [
-0.0869775265455246,
0.029325421899557114,
0.005282561760395765,
-0.06064312160015106,
0.022416407242417336,
-0.08961186558008194,
-0.010202222503721714,
-0.01994055137038231,
0.037726953625679016,
-0.008528468199074268,
0.06531263887882233,
0.010283062234520912,
-0.0038405132945626974,
0.... | -0.004376 |
此渠道忽略回复标签。 ## 确认表情（收到时自动回应） WhatsApp 可以在收到传入消息时立即自动发送表情回应，在机器人生成回复之前。这为用户提供即时反馈，表明他们的消息已收到。 **配置：** ```json { "whatsapp": { "ackReaction": { "emoji": "👀", "direct": true, "group": "mentions" } } } ``` **选项：** - `emoji`（字符串）：用于确认的表情（例如"👀"、"✅"、"📨"）。为空或省略 = 功能禁用。 - `direct`（布尔值，默认：`true`）：在直接/私信聊天中发送表情回应。 - `group`（字符串... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/whatsapp.md | main | opebclaw | [
-0.09913221001625061,
0.03107708878815174,
0.028128935024142265,
-0.00865066796541214,
0.032505374401807785,
-0.027491172775626183,
0.061617009341716766,
-0.0031888424418866634,
0.09568477421998978,
-0.04445965960621834,
0.13897225260734558,
-0.06520836055278778,
0.06890463829040527,
0.041... | 0.050348 |
# Matrix（插件） Matrix 是一个开放的去中心化消息协议。OpenClaw 以 Matrix \*\*用户\*\*身份连接到任意主服务器，因此你需要为机器人创建一个 Matrix 账户。登录后，你可以直接私信机器人或邀请它加入房间（Matrix"群组"）。Beeper 也是一个有效的客户端选项，但它需要启用 E2EE。 状态：通过插件（@vector-im/matrix-bot-sdk）支持。支持私信、房间、话题、媒体、表情回应、投票（发送 + poll-start 作为文本）、位置和 E2EE（需要加密支持）。 ## 需要插件 Matrix 作为插件提供，不包含在核心安装中。 通过 CLI 安装（npm 仓库）： ``... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/matrix.md | main | opebclaw | [
-0.05458226799964905,
0.019312329590320587,
-0.09524965286254883,
-0.06125033646821976,
-0.01667315512895584,
-0.017886072397232056,
0.02278905175626278,
-0.03728541359305382,
0.004783345386385918,
0.02048587240278721,
0.17480263113975525,
-0.04189644753932953,
0.08894869685173035,
-0.0362... | 0.103464 |
ID/别名。 - `channels.matrix.actions`：每个操作的工具限制（reactions/messages/pins/memberInfo/channelInfo）。 ```` | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/matrix.md | main | opebclaw | [
-0.03714686259627342,
-0.04665983468294144,
-0.05073254182934761,
0.008943785913288593,
-0.015723999589681625,
0.013326494954526424,
0.14667464792728424,
-0.018349841237068176,
0.0602111890912056,
0.009707379154860973,
0.14794892072677612,
-0.030564501881599426,
-0.01396532915532589,
0.036... | 0.154961 |
# Nostr \*\*状态：\*\* 可选插件（默认禁用）。 Nostr 是一个去中心化的社交网络协议。此渠道使 OpenClaw 能够通过 NIP-04 接收和回复加密私信（DMs）。 ## 安装（按需） ### 新手引导（推荐） - 新手引导向导（`openclaw onboard`）和 `openclaw channels add` 会列出可选的渠道插件。 - 选择 Nostr 会提示你按需安装插件。 安装默认值： - \*\*Dev 渠道 + git checkout 可用：\*\* 使用本地插件路径。 - \*\*Stable/Beta：\*\* 从 npm 下载。 你可以随时在提示中覆盖选择。 ### 手动安装 ```... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/nostr.md | main | opebclaw | [
0.0008353333105333149,
0.047700054943561554,
-0.043263595551252365,
-0.04425404593348503,
0.018606863915920258,
-0.09948352724313736,
0.026261424645781517,
-0.021690167486667633,
-0.008944560773670673,
0.010732287541031837,
0.10210135579109192,
-0.029429400339722633,
0.021936699748039246,
... | 0.114149 |
# iMessage (imsg) 状态：外部 CLI 集成。Gateway 网关生成 `imsg rpc`（基于 stdio 的 JSON-RPC）。 ## 快速设置（新手） 1. 确保在此 Mac 上已登录"信息"。 2. 安装 `imsg`： - `brew install steipete/tap/imsg` 3. 配置 OpenClaw 的 `channels.imessage.cliPath` 和 `channels.imessage.dbPath`。 4. 启动 Gateway 网关并批准所有 macOS 提示（自动化 + 完全磁盘访问权限）。 最小配置： ```json5 { channels: { imessage... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/imessage.md | main | opebclaw | [
-0.036521147936582565,
-0.020900532603263855,
0.05748682841658592,
-0.039009757339954376,
0.027534177526831627,
-0.12009194493293762,
0.05326768010854721,
0.009257234632968903,
0.061701927334070206,
-0.012826720252633095,
0.061148468405008316,
-0.023008357733488083,
-0.018737105652689934,
... | 0.099256 |
`agents.list[].groupChat.mentionPatterns` 上设置每个智能体的模式。 ## 工作原理（行为） - `imsg` 流式传输消息事件；Gateway 网关将它们规范化为共享渠道信封。 - 回复始终路由回相同的 chat id 或 handle。 ## 类群组会话（`is_group=false`） 某些 iMessage 会话可能有多个参与者，但根据"信息"存储聊天标识符的方式，仍以 `is_group=false` 到达。 如果你在 `channels.imessage.groups` 下显式配置了 `chat_id`，OpenClaw 会将该会话视为"群组"用于： - 会话隔离（独立的 `ag... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/imessage.md | main | opebclaw | [
-0.04324248433113098,
0.0355326384305954,
-0.024278366938233376,
0.01786423847079277,
-0.007707154843956232,
-0.050270192325115204,
0.06572723388671875,
-0.053411807864904404,
0.05225140601396561,
-0.016269151121377945,
0.098880834877491,
-0.04060370847582817,
-0.00006314285565167665,
0.05... | 0.091916 |
# Zalo Personal（非官方） 状态：实验性。此集成通过 `zca-cli` 自动化\*\*个人 Zalo 账户\*\*。 > \*\*警告：\*\*这是一个非官方集成，可能导致账户被暂停/封禁。使用风险自负。 ## 需要插件 Zalo Personal 作为插件提供，不包含在核心安装中。 - 通过 CLI 安装：`openclaw plugins install @openclaw/zalouser` - 或从源码检出安装：`openclaw plugins install ./extensions/zalouser` - 详情：[插件](/plugin) ## 前置条件：zca-cli Gateway 网关机器必须在 ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/zalouser.md | main | opebclaw | [
-0.08899244666099548,
0.048198070377111435,
-0.09478305280208588,
-0.02434282749891281,
-0.0020027717109769583,
-0.08568819612264633,
0.044844139367341995,
-0.005017584189772606,
-0.005208945833146572,
-0.013688287697732449,
0.09876495599746704,
-0.022510956972837448,
0.09105231612920761,
... | 0.072979 |
# Telegram（Bot API） 状态：通过 grammY 支持机器人私信和群组，已可用于生产环境。默认使用长轮询；webhook 可选。 ## 快速设置（入门） 1. 通过 \*\*@BotFather\*\*（[直达链接](https://t.me/BotFather)）创建机器人。确认用户名确实是 `@BotFather`，然后复制 token。 2. 设置 token： - 环境变量：`TELEGRAM\_BOT\_TOKEN=...` - 或配置：`channels.telegram.botToken: "..."`。 - 如果两者都设置了，配置优先（环境变量回退仅适用于默认账户）。 3. 启动 Gateway 网关... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/telegram.md | main | opebclaw | [
-0.06297705322504044,
-0.00283131399191916,
0.04020120948553085,
-0.046199023723602295,
0.02311158925294876,
-0.07129303365945816,
0.012694315053522587,
0.0012494762195274234,
0.018730614334344864,
0.014559333212673664,
0.042847760021686554,
-0.07337111979722977,
0.022033412009477615,
0.02... | 0.123537 |
附加到 Telegram 群组会话键，使每个话题隔离。 - 发送输入指示器和回复时带上 `message\_thread\_id`，使响应保持在话题内。 - 通用话题（线程 id `1`）是特殊的：消息发送省略 `message\_thread\_id`（Telegram 会拒绝），但输入指示器仍然包含它。 - 在模板上下文中暴露 `MessageThreadId` + `IsForum` 用于路由/模板。 - 话题特定配置可在 `channels.telegram.groups..topics.` 下设置（skills、允许列表、自动回复、系统提示、禁用）。 - 话题配置继承群组设置（requireMention、允许列表、ski... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/telegram.md | main | opebclaw | [
-0.012672760523855686,
0.026096707209944725,
-0.019092263653874397,
0.0009311101748608053,
0.01352081447839737,
-0.009728810749948025,
0.07077232748270035,
-0.043714337050914764,
0.010964740999042988,
0.011842046864330769,
0.15444567799568176,
-0.09924441576004028,
0.08771902322769165,
-0.... | 0.114047 |
limit: 5, } ``` 返回缓存中匹配的贴纸： ```json5 { ok: true, count: 2, stickers: [ { fileId: "CAACAgIAAxkBAAI...", emoji: "👋", description: "一只卡通猫热情地挥手", setName: "CoolCats", }, ], } ``` 搜索在描述文本、表情符号字符和集合名称之间使用模糊匹配。 **带线程的示例：** ```json5 { action: "sticker", channel: "telegram", to: "-1001234567890", fileId: "CAACAgIAAxkBAAI...", ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/telegram.md | main | opebclaw | [
-0.10607681423425674,
0.04737183079123497,
0.022514261305332184,
-0.008795814588665962,
0.04287314787507057,
-0.08365683257579803,
0.06020142883062363,
-0.010737680830061436,
0.06083422154188156,
0.00033648806856945157,
0.0876639261841774,
-0.06728342175483704,
0.09103655815124512,
0.03753... | 0.109377 |
时必需）。 - `channels.telegram.webhookPath`：本地 webhook 路径（默认 `/telegram-webhook`）。 - `channels.telegram.actions.reactions`：门控 Telegram 工具反应。 - `channels.telegram.actions.sendMessage`：门控 Telegram 工具消息发送。 - `channels.telegram.actions.deleteMessage`：门控 Telegram 工具消息删除。 - `channels.telegram.actions.sticker`：门控 Telegram 贴纸动作 — ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/telegram.md | main | opebclaw | [
-0.02428712137043476,
-0.0050885374657809734,
-0.009539550170302391,
-0.014187319204211235,
0.05350427329540253,
-0.06171328201889992,
0.10204111039638519,
0.023254750296473503,
0.04744406044483185,
0.04315568134188652,
0.1082790195941925,
-0.06506633013486862,
0.04865983873605728,
0.01081... | 0.175665 |
# Nextcloud Talk（插件） 状态：通过插件支持（webhook 机器人）。支持私信、房间、表情回应和 Markdown 消息。 ## 需要插件 Nextcloud Talk 以插件形式提供，不包含在核心安装包中。 通过 CLI 安装（npm 仓库）： ```bash openclaw plugins install @openclaw/nextcloud-talk ``` 本地检出安装（从 git 仓库运行时）： ```bash openclaw plugins install ./extensions/nextcloud-talk ``` 如果你在配置/新手引导过程中选择了 Nextcloud Talk，并且检测到 ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/nextcloud-talk.md | main | opebclaw | [
-0.033414117991924286,
-0.0018592507112771273,
-0.008411460556089878,
-0.10335829854011536,
0.022295668721199036,
-0.04549219831824303,
0.019550731405615807,
-0.047138359397649765,
0.10090719163417816,
-0.005177399143576622,
0.07564398646354675,
-0.030681071802973747,
0.05065104365348816,
... | 0.050641 |
# Google Chat（Chat API） 状态：已支持通过 Google Chat API webhooks（仅 HTTP）使用私信和空间。 ## 快速设置（新手） 1. 创建一个 Google Cloud 项目并启用 \*\*Google Chat API\*\*。 - 前往：[Google Chat API Credentials](https://console.cloud.google.com/apis/api/chat.googleapis.com/credentials) - 如果 API 尚未启用，请启用它。 2. 创建一个\*\*服务账号\*\*： - 点击 \*\*Create Credentials\*\*... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/googlechat.md | main | opebclaw | [
-0.11667519062757492,
0.019671576097607613,
0.046050067991018295,
-0.09941955655813217,
-0.0206744521856308,
-0.07741441577672958,
0.04215605929493904,
-0.06111569702625275,
0.03555687144398689,
0.01844933070242405,
0.10654504597187042,
-0.08586160838603973,
0.023502664640545845,
-0.010905... | 0.050902 |
users: ["users/1234567890"], systemPrompt: "Short answers only.", }, }, actions: { reactions: true }, typingIndicator: "message", mediaMaxMb: 20, }, }, } ``` 注意事项： - 服务账号凭证也可以通过 `serviceAccount`（JSON 字符串）内联传递。 - 如果未设置 `webhookPath`，默认 webhook 路径为 `/googlechat`。 - 当 `actions.reactions` 启用时，可通过 `reactions` 工具和 `channels ... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/googlechat.md | main | opebclaw | [
-0.04949353262782097,
-0.00032742362236604095,
-0.000060460937675088644,
-0.03526972234249115,
0.020048286765813828,
-0.08212147653102875,
0.04047231748700142,
-0.018765084445476532,
0.03609609976410866,
0.029921486973762512,
0.11753055453300476,
-0.06340359896421432,
0.10398083180189133,
... | 0.06244 |
# Mattermost（插件） 状态：通过插件支持（bot token + WebSocket 事件）。支持频道、群组和私信。 Mattermost 是一个可自托管的团队消息平台；有关产品详情和下载，请访问官方网站 [mattermost.com](https://mattermost.com)。 ## 需要插件 Mattermost 以插件形式提供，不包含在核心安装中。 通过 CLI 安装（npm 注册表）： ```bash openclaw plugins install @openclaw/mattermost ``` 本地检出（从 git 仓库运行时）： ```bash openclaw plugins install .... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/mattermost.md | main | opebclaw | [
-0.06064406782388687,
-0.040541872382164,
-0.06044917553663254,
0.013413215056061745,
0.03485840931534767,
-0.10893848538398743,
0.02114906720817089,
0.015306664630770683,
0.032730165868997574,
-0.005234722979366779,
0.0765136256814003,
-0.07546836882829666,
-0.007875557988882065,
-0.00107... | 0.134364 |
# Discord（Bot API） 状态：已支持通过官方 Discord 机器人网关进行私信和服务器文字频道通信。 ## 快速设置（新手） 1. 创建 Discord 机器人并复制机器人令牌。 2. 在 Discord 应用设置中启用 \*\*Message Content Intent\*\*（如果你计划使用允许列表或名称查找，还需启用 \*\*Server Members Intent\*\*）。 3. 为 OpenClaw 设置令牌： - 环境变量：`DISCORD\_BOT\_TOKEN=...` - 或配置：`channels.discord.token: "..."`。 - 如果两者都设置，配置优先（环境变量回退仅适用... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/discord.md | main | opebclaw | [
-0.031225189566612244,
-0.05515759065747261,
-0.03598807752132416,
-0.020241007208824158,
-0.017210667952895164,
-0.08348271250724792,
0.023841384798288345,
0.014953352510929108,
0.054967958480119705,
-0.01917663961648941,
0.04505909979343414,
-0.06795721501111984,
0.07410050928592682,
-0.... | 0.138504 |
Content Intent**（可能还需要 **Server Members Intent**），然后重启 Gateway 网关。 - **机器人连接但从不在服务器频道回复**： - 缺少 **Message Content Intent**，或 - 机器人缺少频道权限（View/Send/Read History），或 - 你的配置需要提及但你没有提及它，或 - 你的服务器/频道允许列表拒绝了该频道/用户。 - **`requireMention: false` 但仍然没有回复**： - `channels.discord.groupPolicy` 默认为 **allowlist**；将其设置为 `"open"` 或在 `cha... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/discord.md | main | opebclaw | [
0.0295094083994627,
-0.058017946779727936,
-0.03260280564427376,
-0.05959387868642807,
0.01920192502439022,
-0.03749509155750275,
0.02759101614356041,
-0.029741378501057625,
0.04805217683315277,
-0.0017860917141661048,
0.06755702197551727,
-0.04405442625284195,
0.03486764430999756,
0.01258... | 0.067165 |
| 启用 | 上传贴纸 | | polls | 启用 | 创建投票 | | permissions | 启用 | 频道权限快照 | | messages | 启用 | 读取/发送/编辑/删除 | | threads | 启用 | 创建/列出/回复 | | pins | 启用 | 置顶/取消置顶/列出 | | search | 启用 | 消息搜索（预览功能） | | memberInfo | 启用 | 成员信息 | | roleInfo | 启用 | 角色列表 | | channelInfo | 启用 | 频道信息 + 列表 | | channels | 启用 | 频道/类别管理 | | voiceStatus | 启用 | 语音状态... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/discord.md | main | opebclaw | [
-0.01571943797171116,
-0.08739003539085388,
-0.0436166450381279,
-0.05705593526363373,
0.013894332572817802,
0.01554531417787075,
0.040900953114032745,
-0.048219215124845505,
0.02857705019414425,
-0.003271033288910985,
0.08948291838169098,
-0.02191472239792347,
0.06601777672767639,
-0.0484... | 0.116118 |
# LINE（插件） LINE 通过 LINE Messaging API 连接到 OpenClaw。该插件作为 webhook 接收器在 Gateway 网关上运行，使用你的 channel access token + channel secret 进行身份验证。 状态：通过插件支持。支持私信、群聊、媒体、位置、Flex 消息、模板消息和快捷回复。不支持表情回应和话题回复。 ## 需要安装插件 安装 LINE 插件： ```bash openclaw plugins install @openclaw/line ``` 本地检出（从 git 仓库运行时）： ```bash openclaw plugins install ./e... | https://github.com/openclaw/openclaw/blob/main//docs/zh-CN/channels/line.md | main | opebclaw | [
-0.03162166103720665,
0.027690889313817024,
-0.09844566136598587,
-0.02765166573226452,
0.008778013288974762,
-0.10713256150484085,
0.016487272456288338,
0.03976190835237503,
0.03937740996479988,
-0.025078264996409416,
0.07160620391368866,
-0.07920105755329132,
0.03795301541686058,
-0.0402... | 0.080262 |
# SOUL Evil Hook The SOUL Evil hook swaps the \*\*injected\*\* `SOUL.md` content with `SOUL\_EVIL.md` during a purge window or by random chance. It does \*\*not\*\* modify files on disk. ## How It Works When `agent:bootstrap` runs, the hook can replace the `SOUL.md` content in memory before the system prompt is assembl... | https://github.com/openclaw/openclaw/blob/main//docs/hooks/soul-evil.md | main | opebclaw | [
-0.008070236071944237,
0.05934062972664833,
-0.029176494106650352,
0.009857303462922573,
0.03763270750641823,
-0.06521410495042801,
-0.028088903054594994,
-0.05518777668476105,
0.05023223161697388,
-0.0369686521589756,
0.011601785197854042,
0.0619814395904541,
0.02361576445400715,
-0.00809... | 0.077553 |
# Agent Runtime 🤖 OpenClaw runs a single embedded agent runtime derived from \*\*pi-mono\*\*. ## Workspace (required) OpenClaw uses a single agent workspace directory (`agents.defaults.workspace`) as the agent’s \*\*only\*\* working directory (`cwd`) for tools and context. Recommended: use `openclaw setup` to create `... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/agent.md | main | opebclaw | [
-0.03695116564631462,
-0.02919822745025158,
-0.06946897506713867,
0.02629656158387661,
-0.005989986006170511,
-0.058345891535282135,
-0.04874041676521301,
-0.012831129133701324,
0.00904090330004692,
-0.019127914682030678,
0.015913959592580795,
-0.0024495895486325026,
0.03333589434623718,
0... | 0.079619 |
at tool start (no debounce); Control UI streams tool output via agent events when available. More details: [Streaming + chunking](/concepts/streaming). ## Model refs Model refs in config (for example `agents.defaults.model` and `agents.defaults.models`) are parsed by splitting on the \*\*first\*\* `/`. - Use `provider/... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/agent.md | main | opebclaw | [
-0.028276905417442322,
-0.058235540986061096,
-0.0571497306227684,
0.06176113337278366,
0.04183671995997429,
-0.048986468464136124,
0.004124063532799482,
-0.01564992405474186,
0.04864862933754921,
-0.06270891427993774,
0.02523808740079403,
-0.08408153802156448,
-0.01995624043047428,
0.0203... | 0.033043 |
# Usage tracking ## What it is - Pulls provider usage/quota directly from their usage endpoints. - No estimated costs; only the provider-reported windows. ## Where it shows up - `/status` in chats: emoji‑rich status card with session tokens + estimated cost (API key only). Provider usage shows for the \*\*current model... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/usage-tracking.md | main | opebclaw | [
-0.030199773609638214,
0.015738101676106453,
-0.04635290801525116,
0.06425713002681732,
0.0789717361330986,
-0.11807750910520554,
0.0823528841137886,
-0.015086797997355461,
0.09231316298246384,
0.001256767543964088,
0.010230573825538158,
-0.05766257643699646,
-0.02962331846356392,
0.005224... | 0.137645 |
# Group messages (WhatsApp web channel) Goal: let Clawd sit in WhatsApp groups, wake up only when pinged, and keep that thread separate from the personal DM session. Note: `agents.list[].groupChat.mentionPatterns` is now used by Telegram/Discord/Slack/iMessage as well; this doc focuses on WhatsApp-specific behavior. Fo... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/group-messages.md | main | opebclaw | [
-0.054161831736564636,
-0.05878942459821701,
0.03945557028055191,
0.028530582785606384,
-0.010527689941227436,
-0.040072519332170486,
0.10388901829719543,
-0.020494243130087852,
0.0534992478787899,
0.0284743532538414,
0.0406990572810173,
-0.036484211683273315,
0.022546550258994102,
0.08237... | 0.058074 |
`[from: …]` marker so it can address the right person. 4. Session-level directives (`/verbose on`, `/think high`, `/new` or `/reset`, `/compact`) apply only to that group’s session; send them as standalone messages so they register. Your personal DM session remains independent. ## Testing / verification - Manual smoke:... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/group-messages.md | main | opebclaw | [
-0.02547997049987316,
0.03421736881136894,
-0.02708774246275425,
0.010625542141497135,
0.031987112015485764,
-0.10576415061950684,
0.07937516272068024,
-0.07626792788505554,
0.0311275664716959,
0.01590595580637455,
0.05171152949333191,
-0.045549098402261734,
0.0004590791359078139,
-0.00231... | 0.040875 |
# Context Window & Compaction Every model has a \*\*context window\*\* (max tokens it can see). Long-running chats accumulate messages and tool results; once the window is tight, OpenClaw \*\*compacts\*\* older history to stay within limits. ## What compaction is Compaction \*\*summarizes older conversation\*\* into a ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/compaction.md | main | opebclaw | [
-0.07794710248708725,
0.025814114138484,
-0.059190306812524796,
0.050622694194316864,
0.019845062866806984,
-0.05677647516131401,
-0.021039601415395737,
-0.045685701072216034,
0.04774797335267067,
-0.03466728329658508,
-0.0035433403681963682,
0.03405625745654106,
-0.006243275478482246,
0.0... | 0.123434 |
# Gateway architecture Last updated: 2026-01-22 ## Overview - A single long‑lived \*\*Gateway\*\* owns all messaging surfaces (WhatsApp via Baileys, Telegram via grammY, Slack, Discord, Signal, iMessage, WebChat). - Control-plane clients (macOS app, CLI, web UI, automations) connect to the Gateway over \*\*WebSocket\*\... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/architecture.md | main | opebclaw | [
-0.04252318665385246,
-0.023001383990049362,
0.016303332522511482,
-0.03404809907078743,
-0.07065044343471527,
-0.046928588300943375,
0.01892044208943844,
0.006502826232463121,
0.037236593663692474,
0.0834139734506607,
0.013718225061893463,
-0.04003920406103134,
0.040144894272089005,
0.047... | 0.068932 |
WS (also included in `hello-ok`). - Supervision: launchd/systemd for auto‑restart. ## Invariants - Exactly one Gateway controls a single Baileys session per host. - Handshake is mandatory; any non‑JSON or non‑connect first frame is a hard close. - Events are not replayed; clients must refresh on gaps. | https://github.com/openclaw/openclaw/blob/main//docs/concepts/architecture.md | main | opebclaw | [
-0.038590461015701294,
-0.03228231146931648,
-0.0787915363907814,
-0.013182583265006542,
-0.0528821237385273,
0.009316153824329376,
0.03728245943784714,
-0.01360520999878645,
0.005709492601454258,
0.0379367470741272,
-0.035426393151283264,
0.0845162644982338,
0.0018728234572336078,
0.03869... | 0.14617 |
# Messages This page ties together how OpenClaw handles inbound messages, sessions, queueing, streaming, and reasoning visibility. ## Message flow (high level) ``` Inbound message -> routing/bindings -> session key -> queue (if a run is active) -> agent run (streaming + tools) -> outbound replies (channel limits + chun... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/messages.md | main | opebclaw | [
-0.05131746083498001,
-0.08885150402784348,
-0.02273842692375183,
0.02681591920554638,
0.036269668489694595,
-0.10927877575159073,
0.07833217829465866,
-0.05447419360280037,
0.0666506364941597,
0.024126628413796425,
0.030661193653941154,
0.05152108147740364,
-0.0010118200443685055,
-0.0010... | 0.113343 |
(`on|off`, default off) - `agents.defaults.blockStreamingBreak` (`text\_end|message\_end`) - `agents.defaults.blockStreamingChunk` (`minChars|maxChars|breakPreference`) - `agents.defaults.blockStreamingCoalesce` (idle-based batching) - `agents.defaults.humanDelay` (human-like pause between block replies) - Channel over... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/messages.md | main | opebclaw | [
-0.058081094175577164,
-0.0691622942686081,
-0.055397965013980865,
0.005877701099961996,
0.04720737040042877,
-0.048586711287498474,
0.017477158457040787,
0.033209796994924545,
0.0278339684009552,
0.021574262529611588,
-0.016203688457608223,
-0.04855208471417427,
-0.006198681890964508,
0.0... | 0.205262 |
# Channels & routing OpenClaw routes replies \*\*back to the channel where a message came from\*\*. The model does not choose a channel; routing is deterministic and controlled by the host configuration. ## Key terms - \*\*Channel\*\*: `whatsapp`, `telegram`, `discord`, `slack`, `signal`, `imessage`, `webchat`. - \*\*A... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/channel-routing.md | main | opebclaw | [
-0.006627656053751707,
-0.06403552740812302,
-0.035816457122564316,
0.0228001419454813,
-0.037184279412031174,
-0.06532913446426392,
0.06032503396272659,
-0.03035121038556099,
0.07885386049747467,
0.0537397526204586,
0.04168158024549484,
-0.03351535648107529,
0.01461217924952507,
0.0225056... | 0.166926 |
# Retry policy ## Goals - Retry per HTTP request, not per multi-step flow. - Preserve ordering by retrying only the current step. - Avoid duplicating non-idempotent operations. ## Defaults - Attempts: 3 - Max delay cap: 30000 ms - Jitter: 0.1 (10 percent) - Provider defaults: - Telegram min delay: 400 ms - Discord min ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/retry.md | main | opebclaw | [
-0.05946730077266693,
0.04099808633327484,
0.06309134513139725,
0.05064993351697922,
-0.01240120641887188,
-0.0888495072722435,
-0.029939761385321617,
-0.044078487902879715,
0.023916877806186676,
0.03271635249257088,
0.02386467531323433,
0.0017178909620270133,
-0.020695609971880913,
0.0193... | 0.082761 |
# Context “Context” is \*\*everything OpenClaw sends to the model for a run\*\*. It is bounded by the model’s \*\*context window\*\* (token limit). Beginner mental model: - \*\*System prompt\*\* (OpenClaw-built): rules, tools, skills list, time/runtime, and injected workspace files. - \*\*Conversation history\*\*: your... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/context.md | main | opebclaw | [
-0.025986121967434883,
-0.006159087643027306,
-0.08810251206159592,
0.07658515870571136,
0.02337866835296154,
-0.08713832497596741,
0.057662233710289,
0.06774713099002838,
0.01570797525346279,
-0.017624789848923683,
0.0365898534655571,
-0.03306319937109947,
0.056596580892801285,
-0.0239398... | 0.185305 |
- `IDENTITY.md` - `USER.md` - `HEARTBEAT.md` - `BOOTSTRAP.md` (first-run only) Large files are truncated per-file using `agents.defaults.bootstrapMaxChars` (default `20000` chars). `/context` shows \*\*raw vs injected\*\* sizes and whether truncation happened. ## Skills: what’s injected vs loaded on-demand The system p... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/context.md | main | opebclaw | [
-0.03606933727860451,
-0.0004457916656974703,
-0.05990331992506981,
0.011927342973649502,
0.05628957599401474,
-0.06925307959318161,
0.02444514073431492,
0.023181717842817307,
-0.023280059918761253,
0.0043166158720850945,
0.02287360280752182,
0.02131897211074829,
0.0944330245256424,
-0.027... | 0.069616 |
# Session Pruning Session pruning trims \*\*old tool results\*\* from the in-memory context right before each LLM call. It does \*\*not\*\* rewrite the on-disk session history (`\*.jsonl`). ## When it runs - When `mode: "cache-ttl"` is enabled and the last Anthropic call for the session is older than `ttl`. - Only affe... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session-pruning.md | main | opebclaw | [
-0.020441798493266106,
0.052413828670978546,
-0.02602028101682663,
0.04845936968922615,
-0.02230890281498432,
-0.0871206745505333,
-0.05004165321588516,
0.012516818940639496,
0.026990149170160294,
-0.0007675358792766929,
0.0536377914249897,
0.08112175017595291,
-0.04224707558751106,
-0.015... | 0.037074 |
result content cleared]" }` ## Examples Default (off): ```json5 { agent: { contextPruning: { mode: "off" }, }, } ``` Enable TTL-aware pruning: ```json5 { agent: { contextPruning: { mode: "cache-ttl", ttl: "5m" }, }, } ``` Restrict pruning to specific tools: ```json5 { agent: { contextPruning: { mode: "cache-ttl", tools... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session-pruning.md | main | opebclaw | [
-0.019521089270710945,
0.08764774352312088,
-0.03389124944806099,
0.025715570896863937,
-0.0026524034328758717,
-0.03971685469150543,
-0.030000554397702217,
-0.04367382079362869,
0.009480481036007404,
-0.02517484314739704,
0.05857475847005844,
0.07191287726163864,
-0.008615929633378983,
0.... | 0.067121 |
# OAuth OpenClaw supports “subscription auth” via OAuth for providers that offer it (notably \*\*OpenAI Codex (ChatGPT OAuth)\*\*). For Anthropic subscriptions, use the \*\*setup-token\*\* flow. This page explains: - how the OAuth \*\*token exchange\*\* works (PKCE) - where tokens are \*\*stored\*\* (and why) - how to ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/oauth.md | main | opebclaw | [
-0.0776875913143158,
-0.055368583649396896,
-0.05730397626757622,
0.028649702668190002,
-0.0029718673322349787,
-0.11666268855333328,
0.03473934158682823,
0.00293391733430326,
0.04094141721725464,
-0.027921998873353004,
-0.0070346444845199585,
-0.03548862412571907,
0.047587938606739044,
-0... | 0.033185 |
[/concepts/model-failover](/concepts/model-failover) (rotation + cooldown rules) - [/tools/slash-commands](/tools/slash-commands) (command surface) | https://github.com/openclaw/openclaw/blob/main//docs/concepts/oauth.md | main | opebclaw | [
-0.06045961752533913,
-0.06399630755186081,
-0.06125512719154358,
0.020379655063152313,
0.01253601536154747,
0.006548626348376274,
-0.04534171149134636,
0.042539093643426895,
-0.025822103023529053,
0.02521478198468685,
0.12586119771003723,
-0.010712039656937122,
0.016731958836317062,
-0.00... | 0.131773 |
# Agent workspace The workspace is the agent's home. It is the only working directory used for file tools and for workspace context. Keep it private and treat it as memory. This is separate from `~/.openclaw/`, which stores config, credentials, and sessions. \*\*Important:\*\* the workspace is the \*\*default cwd\*\*, ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/agent-workspace.md | main | opebclaw | [
0.0243058018386364,
-0.0349055640399456,
-0.11595216393470764,
0.05545088276267052,
0.022021641954779625,
-0.05029882863163948,
-0.016091063618659973,
-0.03421081602573395,
0.005952651146799326,
-0.04011812433600426,
0.038656800985336304,
-0.0026807638350874186,
0.04259287938475609,
0.0137... | 0.048059 |
is NOT in the workspace These live under `~/.openclaw/` and should NOT be committed to the workspace repo: - `~/.openclaw/openclaw.json` (config) - `~/.openclaw/credentials/` (OAuth tokens, API keys) - `~/.openclaw/agents//sessions/` (session transcripts + metadata) - `~/.openclaw/skills/` (managed skills) If you need ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/agent-workspace.md | main | opebclaw | [
-0.0673665702342987,
-0.08416976034641266,
-0.06222899258136749,
0.04452551528811455,
0.0055349236354231834,
-0.05941975116729736,
-0.016799671575427055,
-0.04078331217169762,
-0.029388107359409332,
-0.004047780763357878,
0.03293174132704735,
0.05982257425785065,
0.03688174858689308,
-0.00... | -0.013391 |
# Multi-Agent Routing Goal: multiple \_isolated\_ agents (separate workspace + `agentDir` + sessions), plus multiple channel accounts (e.g. two WhatsApps) in one running Gateway. Inbound is routed to an agent via bindings. ## What is “one agent”? An \*\*agent\*\* is a fully scoped brain with its own: - \*\*Workspace\*\... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/multi-agent.md | main | opebclaw | [
-0.017617570236325264,
-0.07225333154201508,
-0.10182501375675201,
-0.03642790764570236,
-0.03742794319987297,
-0.017337795346975327,
0.06377391517162323,
-0.0300313550978899,
0.017877351492643356,
-0.017297647893428802,
0.011879605241119862,
-0.0003185365640092641,
0.025784512981772423,
0... | 0.083143 |
fallback to default agent (`agents.list[].default`, else first list entry, default: `main`) ## Multiple accounts / phone numbers Channels that support \*\*multiple accounts\*\* (e.g. WhatsApp) use `accountId` to identify each login. Each `accountId` can be routed to a different agent, so one server can host multiple ph... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/multi-agent.md | main | opebclaw | [
-0.05742321535944939,
-0.03826967626810074,
-0.06927699595689774,
-0.015193554572761059,
-0.06930875778198242,
-0.024203788489103317,
0.06267496198415756,
-0.0054889558814466,
0.05218281224370003,
-0.01362255122512579,
-0.0023723356425762177,
0.009328757412731647,
0.05023423582315445,
0.05... | 0.045705 |
"apply\_patch", "browser", "canvas", "nodes", "cron"], }, }, ], }, bindings: [ { agentId: "family", match: { channel: "whatsapp", peer: { kind: "group", id: "120363999999999999@g.us" }, }, }, ], } ``` Notes: - Tool allow/deny lists are \*\*tools\*\*, not skills. If a skill needs to run a binary, ensure `exec` is allowe... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/multi-agent.md | main | opebclaw | [
-0.04340352863073349,
0.0125059112906456,
-0.04105905070900917,
0.003625450422987342,
0.018378740176558495,
-0.006799777038395405,
0.03324412554502487,
-0.06214780732989311,
-0.08285597711801529,
0.0027635139413177967,
0.041147831827402115,
-0.0096701979637146,
0.019364601001143456,
0.0502... | 0.07018 |
# Timezones OpenClaw standardizes timestamps so the model sees a \*\*single reference time\*\*. ## Message envelopes (local by default) Inbound messages are wrapped in an envelope like: ``` [Provider ... 2026-01-05 16:26 PST] message text ``` The timestamp in the envelope is \*\*host-local by default\*\*, with minutes ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/timezone.md | main | opebclaw | [
0.0123105114325881,
0.04772789403796196,
0.00018453279335517436,
0.04047216475009918,
0.009648412466049194,
-0.09346713125705719,
-0.06481040269136429,
-0.026022067293524742,
0.05188879370689392,
-0.015211161226034164,
-0.016001515090465546,
-0.07559115439653397,
-0.08523315191268921,
0.08... | 0.027545 |
# Streaming + chunking OpenClaw has two separate “streaming” layers: - \*\*Block streaming (channels):\*\* emit completed \*\*blocks\*\* as the assistant writes. These are normal channel messages (not token deltas). - \*\*Token-ish streaming (Telegram only):\*\* update a \*\*draft bubble\*\* with partial text while gen... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/streaming.md | main | opebclaw | [
-0.06237626448273659,
-0.05581201985478401,
0.044648148119449615,
0.018488474190235138,
0.042767856270074844,
-0.12712521851062775,
-0.02680589072406292,
-0.011680214665830135,
0.0792609378695488,
-0.006078875157982111,
-0.0132873160764575,
-0.060146138072013855,
-0.03767869621515274,
-0.0... | 0.161694 |
if very long). - \*\*No block streaming:\*\* `blockStreamingDefault: "off"` (only final reply). \*\*Channel note:\*\* For non-Telegram channels, block streaming is \*\*off unless\*\* `\*.blockStreaming` is explicitly set to `true`. Telegram can stream drafts (`channels.telegram.streamMode`) without block replies. Confi... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/streaming.md | main | opebclaw | [
-0.03481818735599518,
-0.03542788699269295,
0.038437288254499435,
0.008699986152350903,
0.04561394453048706,
-0.06631990522146225,
-0.053236231207847595,
-0.030522305518388748,
0.04053094610571861,
0.06413672864437103,
-0.044790107756853104,
-0.0008420361555181444,
-0.02698509208858013,
0.... | 0.065136 |
# Model providers This page covers \*\*LLM/model providers\*\* (not chat channels like WhatsApp/Telegram). For model selection rules, see [/concepts/models](/concepts/models). ## Quick rules - Model refs use `provider/model` (example: `opencode/claude-opus-4-5`). - If you set `agents.defaults.models`, it becomes the al... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/model-providers.md | main | opebclaw | [
-0.04515964165329933,
-0.05439077317714691,
-0.04426433518528938,
0.018379230052232742,
0.040881309658288956,
-0.02943587861955166,
-0.02425069361925125,
-0.012908130884170532,
0.0213195588439703,
0.025205401703715324,
0.034456975758075714,
-0.06818094104528427,
-0.005670203361660242,
0.07... | 0.121581 |
api: "openai-completions", models: [{ id: "kimi-k2.5", name: "Kimi K2.5" }], }, }, }, } ``` ### Kimi Coding Kimi Coding uses Moonshot AI's Anthropic-compatible endpoint: - Provider: `kimi-coding` - Auth: `KIMI\_API\_KEY` - Example model: `kimi-coding/k2p5` ```json5 { env: { KIMI\_API\_KEY: "sk-..." }, agents: { default... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/model-providers.md | main | opebclaw | [
-0.05135742202401161,
-0.0011503025889396667,
-0.04870936647057533,
0.013950576074421406,
0.04960789531469345,
-0.02767864800989628,
-0.011919884011149406,
-0.03316294774413109,
-0.0003860114957205951,
-0.0010550820734351873,
0.051846571266651154,
-0.08777447789907455,
0.04360524192452431,
... | 0.097434 |
# Typing indicators Typing indicators are sent to the chat channel while a run is active. Use `agents.defaults.typingMode` to control \*\*when\*\* typing starts and `typingIntervalSeconds` to control \*\*how often\*\* it refreshes. ## Defaults When `agents.defaults.typingMode` is \*\*unset\*\*, OpenClaw keeps the legac... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/typing-indicators.md | main | opebclaw | [
-0.054928891360759735,
-0.08551354706287384,
0.0011123532894998789,
0.10623891651630402,
-0.009919523261487484,
-0.015294687822461128,
0.07840283215045929,
-0.01031300239264965,
0.09064920991659164,
-0.03293551132082939,
0.005720982328057289,
-0.04535577818751335,
-0.031074747443199158,
0.... | 0.187907 |
# Model failover OpenClaw handles failures in two stages: 1. \*\*Auth profile rotation\*\* within the current provider. 2. \*\*Model fallback\*\* to the next model in `agents.defaults.model.fallbacks`. This doc explains the runtime rules and the data that backs them. ## Auth storage (keys + OAuth) OpenClaw uses \*\*aut... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/model-failover.md | main | opebclaw | [
-0.035460859537124634,
-0.01048616785556078,
-0.05835700035095215,
0.0814080610871315,
0.0547347217798233,
-0.10420487821102142,
-0.06828140467405319,
0.0062948050908744335,
-0.021641533821821213,
-0.08013992011547089,
0.05402529612183571,
-0.0048457360826432705,
0.08655320852994919,
-0.02... | -0.002955 |
`auth-profiles.json`: ```json { "usageStats": { "provider:profile": { "disabledUntil": 1736178000000, "disabledReason": "billing" } } } ``` Defaults: - Billing backoff starts at \*\*5 hours\*\*, doubles per billing failure, and caps at \*\*24 hours\*\*. - Backoff counters reset if the profile hasn’t failed for \*\*24 h... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/model-failover.md | main | opebclaw | [
-0.05533677712082863,
0.006319382227957249,
-0.056439951062202454,
0.09280730783939362,
0.011940152384340763,
-0.047033969312906265,
-0.028361178934574127,
0.014050199650228024,
-0.024439571425318718,
-0.048805445432662964,
0.03213915973901749,
0.016703126952052116,
0.042273957282304764,
0... | 0.008922 |
# Memory OpenClaw memory is \*\*plain Markdown in the agent workspace\*\*. The files are the source of truth; the model only "remembers" what gets written to disk. Memory search tools are provided by the active memory plugin (default: `memory-core`). Disable memory plugins with `plugins.slots.memory = "none"`. ## Memor... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/memory.md | main | opebclaw | [
0.016076339408755302,
0.021671166643500328,
-0.1188412681221962,
0.11086111515760422,
0.003255744930356741,
0.01407179981470108,
-0.03174273297190666,
0.0026396941393613815,
0.013424628414213657,
0.017045391723513603,
0.02876666933298111,
0.10129180550575256,
0.029580840840935707,
-0.00477... | 0.11528 |
Markdown stays the source of truth; OpenClaw shells out to QMD for retrieval. Key points: \*\*Prereqs\*\* - Disabled by default. Opt in per-config (`memory.backend = "qmd"`). - Install the QMD CLI separately (`bun install -g github.com/tobi/qmd` or grab a release) and make sure the `qmd` binary is on the gateway’s `PAT... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/memory.md | main | opebclaw | [
-0.03590930998325348,
-0.013998620212078094,
-0.0645570382475853,
0.006309633143246174,
-0.01409865077584982,
-0.011763684451580048,
-0.08852588385343552,
0.04089094325900078,
0.013071868568658829,
0.016234328970313072,
0.04100187495350838,
-0.027591601014137268,
0.004599283915013075,
0.00... | 0.05407 |
{ interval: "5m", debounceMs: 15000 }, limits: { maxResults: 6, timeoutMs: 4000 }, scope: { default: "deny", rules: [{ action: "allow", match: { chatType: "direct" } }] }, paths: [ { name: "docs", path: "~/notes", pattern: "\*\*/\*.md" } ] } } ``` \*\*Citations & fallback\*\* - `memory.citations` applies regardless of ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/memory.md | main | opebclaw | [
-0.09444921463727951,
0.03346100077033043,
-0.02251295931637287,
0.034459251910448074,
-0.019644180312752724,
-0.025576554238796234,
-0.06681867688894272,
0.012534497305750847,
0.07641307264566422,
0.014504600316286087,
-0.032020460814237595,
0.041514765471220016,
0.013151509687304497,
0.0... | 0.069785 |
work - `memory\_search` semantically searches Markdown chunks (~400 token target, 80-token overlap) from `MEMORY.md` + `memory/\*\*/\*.md`. It returns snippet text (capped ~700 chars), file path, line range, score, provider/model, and whether we fell back from local → remote embeddings. No full file payload is returned... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/memory.md | main | opebclaw | [
0.009926130063831806,
0.016214948147535324,
-0.10599066317081451,
0.07276918739080429,
-0.002780328970402479,
-0.0031024315394461155,
-0.030844461172819138,
0.028051529079675674,
-0.008790775202214718,
0.006623093970119953,
-0.023742245510220528,
0.08599502593278885,
0.050231847912073135,
... | 0.057889 |
(experimental) You can optionally index \*\*session transcripts\*\* and surface them via `memory\_search`. This is gated behind an experimental flag. ```json5 agents: { defaults: { memorySearch: { experimental: { sessionMemory: true }, sources: ["memory", "sessions"] } } } ``` Notes: - Session indexing is \*\*opt-in\*\... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/memory.md | main | opebclaw | [
-0.05244947597384453,
0.08337228000164032,
-0.0659525915980339,
0.08275305479764938,
0.03982362523674965,
-0.02177453599870205,
0.031266551464796066,
-0.025966420769691467,
0.020203962922096252,
0.02003440074622631,
-0.019294893369078636,
0.05675097182393074,
-0.01879291795194149,
0.005598... | 0.143296 |
# TypeBox as protocol source of truth Last updated: 2026-01-10 TypeBox is a TypeScript-first schema library. We use it to define the \*\*Gateway WebSocket protocol\*\* (handshake, request/response, server events). Those schemas drive \*\*runtime validation\*\*, \*\*JSON Schema export\*\*, and \*\*Swift codegen\*\* for ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/typebox.md | main | opebclaw | [
-0.020540833473205566,
-0.030534442514181137,
-0.001371055725030601,
0.02081719972193241,
-0.009632095694541931,
-0.040377359837293625,
0.0230368934571743,
-0.028555456548929214,
0.008196856826543808,
-0.044449854642152786,
-0.07227936387062073,
-0.03380359709262848,
-0.07757166028022766,
... | 0.177378 |
"connect", params: { minProtocol: 3, maxProtocol: 3, client: { id: "cli", displayName: "example", version: "dev", platform: "node", mode: "cli", }, }, }), ); }); ws.on("message", (data) => { const msg = JSON.parse(String(data)); if (msg.type === "res" && msg.id === "c1" && msg.ok) { ws.send(JSON.stringify({ type: "req"... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/typebox.md | main | opebclaw | [
-0.03537807613611221,
0.067760169506073,
-0.018393460661172867,
0.05112295225262642,
-0.020635880529880524,
-0.0654032751917839,
0.057758599519729614,
0.12123054265975952,
0.04148176312446594,
-0.044168345630168915,
-0.06581293046474457,
-0.010674258694052696,
-0.004896054044365883,
0.0232... | 0.060424 |
# Models CLI See [/concepts/model-failover](/concepts/model-failover) for auth profile rotation, cooldowns, and how that interacts with fallbacks. Quick provider overview + examples: [/concepts/model-providers](/concepts/model-providers). ## How model selection works OpenClaw selects models in this order: 1. \*\*Primar... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/models.md | main | opebclaw | [
-0.02044876106083393,
-0.055748164653778076,
-0.040866587311029434,
0.1002403199672699,
0.08336778730154037,
-0.05121072754263878,
-0.028554392978549004,
0.01055977027863264,
-0.01952320709824562,
-0.043712690472602844,
0.07806717604398727,
0.005736805032938719,
0.08290886133909225,
-0.010... | 0.018045 |
`--all`: full catalog - `--local`: local providers only - `--provider `: filter by provider - `--plain`: one model per line - `--json`: machine‑readable output ### `models status` Shows the resolved primary model, fallbacks, image model, and an auth overview of configured providers. It also surfaces OAuth expiry status... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/models.md | main | opebclaw | [
-0.021547509357333183,
0.029180385172367096,
-0.08088983595371246,
0.02706882357597351,
0.07080481201410294,
-0.0716298371553421,
-0.016388682648539543,
0.02686665765941143,
-0.02749994397163391,
-0.04337291046977043,
0.05907468497753143,
-0.025050748139619827,
0.031111186370253563,
0.0160... | 0.010634 |
# Agent Loop (OpenClaw) An agentic loop is the full “real” run of an agent: intake → context assembly → model inference → tool execution → streaming replies → persistence. It’s the authoritative path that turns a message into actions and a final reply, while keeping session state consistent. In OpenClaw, a loop is a si... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/agent-loop.md | main | opebclaw | [
-0.030445309355854988,
-0.03626398742198944,
-0.15161728858947754,
0.012150431983172894,
0.016231708228588104,
-0.07589345425367355,
0.0013023328501731157,
-0.011461975984275341,
0.061443254351615906,
-0.025808651000261307,
0.017678245902061462,
0.001303599332459271,
0.0012510457308962941,
... | 0.131687 |
`message\_sent`\*\*: inbound + outbound message hooks. - \*\*`session\_start` / `session\_end`\*\*: session lifecycle boundaries. - \*\*`gateway\_start` / `gateway\_stop`\*\*: gateway lifecycle events. See [Plugins](/plugin#plugin-hooks) for the hook API and registration details. ## Streaming + partial replies - Assist... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/agent-loop.md | main | opebclaw | [
-0.06734863668680191,
-0.06633787602186203,
-0.009343307465314865,
0.014608935452997684,
0.05064060911536217,
-0.04701276868581772,
0.0529891774058342,
-0.013660101220011711,
0.056889645755290985,
0.01356162503361702,
-0.02000896818935871,
-0.04843418672680855,
-0.021663906052708626,
0.008... | 0.168117 |
# Markdown formatting OpenClaw formats outbound Markdown by converting it into a shared intermediate representation (IR) before rendering channel-specific output. The IR keeps the source text intact while carrying style/link spans so chunking and rendering can stay consistent across channels. ## Goals - \*\*Consistency... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/markdown-formatting.md | main | opebclaw | [
-0.043157730251550674,
0.03434517979621887,
0.020932752639055252,
0.03226003050804138,
0.036313001066446304,
-0.059807389974594116,
-0.05870551988482475,
0.007411098573356867,
0.0025115865282714367,
0.013752617873251438,
-0.07358933985233307,
-0.03168874233961105,
-0.00822532456368208,
-0.... | 0.016106 |
adapter to use the new chunker and renderer. 5. **Test:** add or update format tests and an outbound delivery test if the channel uses chunking. ## Common gotchas - Slack angle-bracket tokens (`<@U123>`, `<#C123>`, ``) must be preserved; escape raw HTML safely. - Telegram HTML requires escaping text outside tags to avo... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/markdown-formatting.md | main | opebclaw | [
-0.10318917036056519,
0.02043810859322548,
0.019545597955584526,
-0.04444880038499832,
-0.0013414269778877497,
-0.03494350612163544,
-0.03987486660480499,
-0.01855139434337616,
-0.03355179727077484,
0.010069239884614944,
-0.006467525847256184,
-0.03820650652050972,
0.0034937101881951094,
0... | 0.082686 |
# Session Tools Goal: small, hard-to-misuse tool set so agents can list sessions, fetch history, and send to another session. ## Tool Names - `sessions\_list` - `sessions\_history` - `sessions\_send` - `sessions\_spawn` ## Key Model - Main direct chat bucket is always the literal key `"main"` (resolved to the current a... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session-tool.md | main | opebclaw | [
-0.05458036810159683,
0.02673329785466194,
-0.04585045203566551,
0.031648747622966766,
-0.021658165380358696,
-0.03058413602411747,
0.08901402354240417,
0.003398350439965725,
-0.019273247569799423,
0.06283790618181229,
-0.03512084111571312,
-0.06388052552938461,
-0.01046472042798996,
0.046... | 0.063899 |
\*\*agent‑to‑agent announce step\*\* (target agent only): - Reply exactly `ANNOUNCE\_SKIP` to stay silent. - Any other reply is sent to the target channel. - Announce step includes the original request + round‑1 reply + latest ping‑pong reply. ## Channel Field - For groups, `channel` is the channel recorded on the sess... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session-tool.md | main | opebclaw | [
-0.052132852375507355,
0.03203554451465607,
0.007240655831992626,
-0.002470061182975769,
0.018325654789805412,
-0.049746811389923096,
0.10407765209674835,
-0.04865540936589241,
0.05140526592731476,
0.036088977009058,
-0.0245609600096941,
-0.026780573651194572,
-0.05104507878422737,
0.03604... | 0.105934 |
# Command Queue (2026-01-16) We serialize inbound auto-reply runs (all channels) through a tiny in-process queue to prevent multiple agent runs from colliding, while still allowing safe parallelism across sessions. ## Why - Auto-reply runs can be expensive (LLM calls) and can collide when multiple inbound messages arri... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/queue.md | main | opebclaw | [
-0.046516239643096924,
-0.04851388558745384,
-0.06493265926837921,
-0.044298119843006134,
-0.10358322411775589,
-0.04507958143949509,
0.045204970985651016,
-0.005792254116386175,
0.040358442813158035,
0.06446926295757294,
-0.01015609223395586,
-0.005247548688203096,
-0.005258535034954548,
... | 0.172262 |
inbound replies. - Per-session lanes guarantee that only one agent run touches a given session at a time. - No external dependencies or background worker threads; pure TypeScript + promises. ## Troubleshooting - If commands seem stuck, enable verbose logs and look for “queued for …ms” lines to confirm the queue is drai... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/queue.md | main | opebclaw | [
-0.04188026115298271,
-0.05372555926442146,
-0.024338075891137123,
0.03158187493681908,
-0.018976736813783646,
-0.05240830406546593,
0.03237620368599892,
-0.03546309098601341,
0.03767430782318115,
0.05175169184803963,
-0.08226443827152252,
-0.00730778556317091,
-0.1073305755853653,
0.03104... | 0.078789 |
## Highlights WhatsApp, Telegram, Discord, and iMessage with a single Gateway. Add Mattermost and more with extensions. Multi-agent routing with isolated sessions. Images, audio, and documents in and out. Web Control UI and macOS companion app. iOS and Android nodes with Canvas support. ## Full list - WhatsApp integrat... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/features.md | main | opebclaw | [
-0.05689702928066254,
-0.04868466779589653,
0.02826215699315071,
-0.03972471505403519,
0.06448648124933243,
-0.06217474862933159,
-0.004464621655642986,
0.03881975635886192,
-0.0025156689807772636,
0.06786975264549255,
0.04155704006552696,
-0.030323978513479233,
0.009764498099684715,
0.086... | 0.160278 |
# Session Management OpenClaw treats \*\*one direct-chat session per agent\*\* as primary. Direct chats collapse to `agent::` (default `main`), while group/channel chats get their own keys. `session.mainKey` is honored. Use `session.dmScope` to control how \*\*direct messages\*\* are grouped: - `main` (default): all DM... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session.md | main | opebclaw | [
-0.015107673592865467,
-0.06035982444882393,
-0.053169701248407364,
0.023119991645216942,
-0.031057875603437424,
-0.10282140970230103,
0.10587683320045471,
-0.0012724230764433742,
0.015038037672638893,
0.03712598234415054,
0.02318589948117733,
-0.019976967945694923,
0.037666141986846924,
0... | 0.070323 |
isolation. - Legacy `group:` keys are still recognized for migration. - Inbound contexts may still use `group:`; the channel is inferred from `Provider` and normalized to the canonical `agent:::group:` form. - Other sources: - Cron jobs: `cron:` - Webhooks: `hook:` (unless explicitly set by the hook) - Node runs: `node... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session.md | main | opebclaw | [
-0.09434538334608078,
-0.016932260245084763,
-0.06366483122110367,
0.06353851407766342,
0.07757102698087692,
-0.08381055295467377,
0.024660471826791763,
-0.044062793254852295,
0.010929071344435215,
-0.0272297915071249,
0.04756796360015869,
0.023921774700284004,
-0.024606704711914062,
-0.02... | 0.115392 |
is used, current thinking/verbose toggles, and when your WhatsApp web creds were last refreshed (helps spot relink needs). - Send `/context list` or `/context detail` to see what’s in the system prompt and injected workspace files (and the biggest context contributors). - Send `/stop` as a standalone message to abort t... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/session.md | main | opebclaw | [
-0.060120224952697754,
0.03149251267313957,
0.012121928855776787,
-0.007135417778044939,
0.04984812065958977,
-0.029775377362966537,
0.054926369339227676,
-0.0019919821061193943,
0.041623082011938095,
0.060311831533908844,
-0.011010491289198399,
0.02790595032274723,
-0.022235440090298653,
... | 0.113085 |
# System Prompt OpenClaw builds a custom system prompt for every agent run. The prompt is \*\*OpenClaw-owned\*\* and does not use the p-coding-agent default prompt. The prompt is assembled by OpenClaw and injected into each agent run. ## Structure The prompt is intentionally compact and uses fixed sections: - \*\*Tooli... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/system-prompt.md | main | opebclaw | [
-0.006138477474451065,
-0.006984990555793047,
-0.08940403163433075,
0.04148362576961517,
0.019699793308973312,
-0.043569695204496384,
-0.00956397969275713,
-0.011275921948254108,
-0.03292009234428406,
0.01950230821967125,
0.016057435423135757,
-0.024217823520302773,
0.04875919595360756,
-0... | 0.124005 |
section is omitted. ``` ... ... ... ``` This keeps the base prompt small while still enabling targeted skill usage. ## Documentation When available, the system prompt includes a \*\*Documentation\*\* section that points to the local OpenClaw docs directory (either `docs/` in the repo workspace or the bundled npm packag... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/system-prompt.md | main | opebclaw | [
-0.0022466848604381084,
0.02737252041697502,
-0.0374993272125721,
0.06056595966219902,
0.04886868968605995,
-0.08844122290611267,
-0.04271445795893669,
-0.00702266488224268,
-0.06327050179243088,
-0.011869595386087894,
0.0046590655110776424,
-0.017029311507940292,
0.0421333946287632,
-0.02... | 0.057181 |
# Groups OpenClaw treats group chats consistently across surfaces: WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Microsoft Teams. ## Beginner intro (2 minutes) OpenClaw “lives” on your own messaging accounts. There is no separate WhatsApp bot user. If \*\*you\*\* are in a group, OpenClaw can see that group and ... | https://github.com/openclaw/openclaw/blob/main//docs/concepts/groups.md | main | opebclaw | [
-0.09113698452711105,
-0.04864346608519554,
-0.0008001552196219563,
0.06957617402076721,
0.039414726197719574,
-0.1046954095363617,
0.03742660954594612,
-0.0510735809803009,
0.05436762794852257,
0.009101013652980328,
0.04059239849448204,
-0.06838351488113403,
-0.001633316627703607,
0.03771... | 0.041375 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.