菜单
OpenProse workflow language for OpenClaw. Use when running the /prose slash command or the open-prose plugin. Canonical docs: docs.openclaw.ai/prose. Triggers on: prose, openprose, /prose, .prose files, prose workflow, multi-agent workflow, parallel agents, agent orchestration.
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
OpenClaw channel configuration. Use when setting up, auditing, or troubleshooting messaging channels: Discord, Telegram, WhatsApp, Slack, Signal, Matrix, iMessage, IRC, Feishu, LINE, GoogleChat, Mattermost, Microsoft Teams, Nextcloud Talk, Nostr, QQ Bot, Synology Chat, Twitch, Tlon, Zalo, Voice Call, Raft, ClickClack, plus location and qa-channel. (BlueBubbles was the prior iMessage path and is now removed; see references/bluebubbles.md for the migration guide.) Triggers on: channel config, dmPolicy, group policy, allowFrom, pairing, broadcast groups, channel pairing, matrix migration, push rules, raft consensus, location, setup channel, connect channel.
OpenClaw configuration expert and system doctor. Use for unclassified config questions that don't map to a more specific skill — for specific areas use their dedicated skills (see skill map below). Triggers on: openclaw expert, config doctor, system health, gateway diagnostics, openclaw master, openclaw overview, openclaw help.
OpenClaw model providers. Use when configuring LLM providers (60+ individual provider files tracked: OpenAI, Anthropic, Gemini, Bedrock, Ollama, DeepSeek, Groq, Together, Cerebras, xAI, Cohere, Deepgram, SenseAudio, Tencent, VolcEngine, and more), API keys, OAuth, auth profiles, model failover, API key rotation, billing disable backoff, and usage tracking. Triggers on: provider, model, API key, OAuth, Anthropic, OpenAI, Gemini, Bedrock, Ollama, failover, auth profile, billing disable, key rotation, Cohere, Deepgram, transcription, SenseAudio, speech to text, Tencent, VolcEngine, Doubao, voice provider, STT provider.
Guided first-run setup for ClawDoc (the docs/skills themselves), NOT for OpenClaw runtime first-run (use openclaw-start for that). Walks through installing ClawDoc skills, configuring docs sync, and end-to-end verification of the ClawDoc layer. Triggers on: "ClawDoc first run", "ClawDoc setup", "install ClawDoc", "ClawDoc getting started", "set up ClawDoc skills", "ClawDoc install verification".
OpenClaw multi-agent system. Use when setting up multiple agents, bindings, channel routing, per-agent sandbox, tool policies, ACP runtimes, workspace bootstrap, subagent limits, context pruning, or per-agent thinking/reasoning overrides. Also covers: multi-agent-sandbox-tools, qa-e2e-automation, session-tool reference docs. Triggers on: "multi-agent", "bindings", "routing", "agentId", "workspace", "sandbox", "tool policy", "per-agent", "ACP", "acp", "runtime", "bootstrap", "subagent", "sessions_spawn", "contextPruning", "thinkingDefault", "reasoningDefault", "fastModeDefault", "execApprovals", "sandbox mode", "sandbox scope".
Automation in OpenClaw — cron jobs, hooks, tasks, webhooks, poll, standing orders, Task Flow, Clawflow. Also covers: auth-monitoring, clawflow, commitments, cron-vs-heartbeat, standing-orders reference docs. Use when: scheduling recurring tasks, setting up cron jobs, configuring webhooks, reacting to events with hooks, running background tasks, orchestrating multi-step workflows, using standing orders to inject persistent instructions. Triggers on: "automation", "cron", "schedule", "webhook", "hook", "task", "background", "standing order", "recurring", "every x minutes", "task flow", "clawflow", "poll", "CI", "logging", "automation troubleshooting", "gmail", "pubsub", "gmail pubsub", "cron vs heartbeat", "scheduled task", "commitments", "auth monitoring", "clawflow workflow".
| name | openclaw-prose |
| description | OpenProse workflow language for OpenClaw. Use when running the /prose slash command or the open-prose plugin. Canonical docs: docs.openclaw.ai/prose. Triggers on: prose, openprose, /prose, .prose files, prose workflow, multi-agent workflow, parallel agents, agent orchestration. |
OpenProse is a portable, markdown-first workflow format for orchestrating AI sessions. In OpenClaw it ships as a plugin (open-prose) that installs a skill pack and a /prose slash command. Programs live in .prose files and can spawn multiple sub-agents with explicit control flow.
# Enable the plugin (bundled, disabled by default) and restart the gateway
openclaw plugins enable open-prose
openclaw gateway restart
# Run a program
/prose run ./research.prose
The canonical OpenProse documentation lives in upstream OpenClaw:
openclaw/docs/prose.mdopenclaw/docs/plugins/reference/open-prose.mdThis skill is a thin pointer. The bundled reference docs that previously lived at references/ (install, slash-command, programming, runtime-mapping, state, examples) were dropped in v1.7.10 because their content was not verifiable against the upstream source — the canonical material is in the upstream docs linked above. If you need content that is not yet covered upstream, see the upstream source files directly or open an issue against the ClawDoc repo.