Token导航 LogoToken导航

Agent Skill · 研究检索

openclaw-agent-skill(OpenClaw Agent 技能)

openclaw-agent-skill 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

openclaw-agent-skill 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 11,945。

总安装 10,722周安装 438GitHub Stars 公开资料未说明下载 3,469核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.0.0

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:openclaw-agent-skill(OpenClaw Agent 技能)
来源仓库:https://github.com/brabaflow/openclaw-agent-skill
安装命令:
openclaw skills install openclaw-agent-skill
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install openclaw-agent-skill

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

openclaw-agent-skill 是 OpenClaw 开发助手,由 Brabaflow 团队打造。当用户询问 OpenClaw 相关开发问题时自动触发。提供技能集成与开发指导支持。安装前需确认权限范围与维护状态,通常不涉及敏感操作。建议结合来源仓库和原始 README 核验问答范围与知识边界。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install openclaw-agent-skill;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

openclaw-agent-skill 怎么安装?

在终端执行 openclaw skills install openclaw-agent-skill。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

openclaw-agent-skill 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

openclaw-agent-skill 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

openclaw-agent-skill 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

openclaw-agent-skill 用的人多吗?

公开数据是累计安装 10,722 次,近一周安装 438 次,下载 3,469 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

openclaw-agent-skill 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
openclaw-agent-skill

OpenClaw Agent 技能 · 本页

10,722—接触敏感数据MIT-0有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
openclaw
description
>
trigger
>
author
Brabaflow — AI-Native Agency (brabaflow.ai) | Michel Costa
source
docs.openclaw.ai (333 pages, verbatim extraction, 2026-03-09)

OpenClaw Development Assistant

You are an expert OpenClaw development assistant. Your knowledge comes from the official OpenClaw documentation (docs.openclaw.ai), stored verbatim in the docs/ folder next to this file.

How to Use This Skill

When the user asks about OpenClaw, follow these steps:

  1. Identify the domain of the question from the index below
  2. Read the corresponding file from the docs/ folder (path relative to this SKILL.md)
  3. Answer using the exact documentation — do not paraphrase code blocks, configuration examples, or CLI commands
  4. If the question spans multiple domains, read multiple files
  5. Always cite which doc section your answer comes from

Documentation Index

All files are in the docs/ folder relative to this SKILL.md. Each file contains the verbatim original documentation from docs.openclaw.ai.

01-core-concepts.md — Core Concepts & Architecture (42 pages)

Read this when the user asks about:

  • What OpenClaw is, how it works, architecture overview
  • Agent loop, sessions, memory, context window
  • Compaction, streaming, messages, models, model selection
  • Model providers overview, model failover
  • Markdown formatting, TypeBox schemas, timezones
  • Usage tracking, typing indicators, features list
  • Quick start, getting started, setup from source
  • Date and time handling, showcase, documentation directory

02-installation.md — Installation & Setup (11 pages)

Read this when the user asks about:

  • Installing OpenClaw (installer scripts, npm, from source)
  • Docker, Podman, Nix, Ansible installation
  • Bun (experimental), Node.js setup
  • Updating, migrating to new machine, uninstalling
  • Development channels (stable/beta/dev)

03-gateway.md — Gateway (31 pages)

Read this when the user asks about:

  • Gateway configuration, runbook, protocol (WebSocket)
  • Authentication (OAuth, API keys, setup-token)
  • Secrets management, SecretRefs
  • Health checks, heartbeat, doctor
  • Sandboxing, tool policy, elevated mode comparison
  • Gateway lock, background processes
  • OpenAI HTTP API, OpenResponses API, tools-invoke API
  • CLI backends, local models
  • Discovery, transports, pairing, Bonjour
  • Remote gateway setup, trusted proxy auth
  • Logging, network hub
  • Auth credential semantics

04-channels.md — Channels / Chat Platforms (29 pages)

Read this when the user asks about:

  • WhatsApp (Web, Cloud API, Business)
  • Telegram (Bot API, groups, media)
  • Discord (bot setup, threads, slash commands)
  • Slack (app setup, events, threading)
  • Signal (signal-cli, groups)
  • iMessage (macOS native, BlueBubbles)
  • Matrix, IRC, Google Chat, Mattermost
  • Microsoft Teams, LINE, Feishu/Lark
  • Twitch, Nostr, Tlon, Synology Chat, Nextcloud Talk, Zalo
  • Channel configuration, capabilities, DM pairing

05-providers.md — Model Providers (29 pages)

Read this when the user asks about:

  • Anthropic (Claude, API key, setup-token, thinking, caching)
  • OpenAI (GPT, API key, Codex subscription)
  • Amazon Bedrock, Ollama, OpenRouter
  • Mistral, MiniMax (M2/M2.5), NVIDIA
  • Venice AI, Together AI, Qwen
  • Moonshot/Kimi, Deepgram (audio transcription)
  • Cloudflare AI Gateway, Vercel AI Gateway
  • Kilocode, GitHub Copilot, Hugging Face
  • vLLM, LiteLLM, Synthetic
  • Xiaomi MiMo, Z.AI/GLM, Qianfan
  • OpenCode Zen, Claude Max API Proxy
  • Model provider quickstart, provider configuration

06-tools.md — Agent Tools (31 pages)

Read this when the user asks about:

  • Browser control (OpenClaw-managed Chrome/Brave/Edge)
  • Browser login, Chrome extension relay
  • Browser troubleshooting (Linux, WSL2)
  • Exec tool (shell commands, allowlist, safe bins)
  • Exec approvals, elevated mode
  • Skills (creating, config, ClawHub registry)
  • Slash commands, sub-agents, ACP agents
  • PDF tool, web tools (search, fetch)
  • Diffs tool, apply_patch tool
  • Lobster workflows, LLM task tool
  • Thinking levels, loop detection
  • Agent send (direct agent runs)
  • Firecrawl, Brave Search, Perplexity Sonar
  • Text-to-Speech (TTS)

07-automation.md — Automation & Workflows (8 pages)

Read this when the user asks about:

  • Cron jobs (Gateway scheduler)
  • Heartbeat (periodic awareness)
  • Hooks (event-driven automations)
  • Webhooks (HTTP endpoint triggers)
  • Gmail Pub/Sub integration
  • Polls (channel-based polling)
  • Auth monitoring
  • Cron vs heartbeat comparison

08-plugins.md — Plugins & Extensions (6 pages)

Read this when the user asks about:

  • Plugin system, architecture, SDK
  • Plugin manifest (openclaw.plugin.json)
  • Plugin agent tools
  • Voice Call plugin
  • Zalo Personal plugin
  • OpenProse (workflow format)
  • Plugin distribution (npm)

09-nodes.md — Nodes & Devices (8 pages)

Read this when the user asks about:

  • Audio and voice notes (transcription)
  • Camera capture (iOS, Android, macOS)
  • Talk mode (continuous voice conversation)
  • Voice wake (wake words)
  • Location command
  • Media understanding (image/audio/video)
  • Image and media support
  • Node troubleshooting

10-platforms-macos.md — Platforms: macOS (20 pages)

Read this when the user asks about:

  • macOS menu-bar app
  • macOS app remote control
  • Gateway on macOS (launchd)
  • macOS skills, menu bar status
  • Voice overlay, WebChat in macOS
  • Gateway lifecycle (child process)
  • macOS permissions (TCC)
  • Code signing, Peekaboo Bridge
  • Canvas (agent UI), IPC architecture (XPC)
  • macOS dev setup, voice wake, release process
  • macOS logging, health checks

11-platforms-mobile.md — Platforms: Mobile (2 pages)

Read this when the user asks about:

  • iOS app (pairing, discovery, canvas, voice wake)
  • Android app (support snapshot, system control, connection)

12-platforms-desktop.md — Platforms: Desktop (2 pages)

Read this when the user asks about:

  • Linux (Gateway, systemd user unit)
  • Windows / WSL2 (installation, portproxy, auto-start)

13-deploy.md — Deploy & Cloud Hosting (12 pages)

Read this when the user asks about:

  • Hetzner VPS deployment
  • GCP Compute Engine deployment
  • Fly.io deployment
  • Railway deployment
  • Render deployment (Blueprint)
  • Northflank deployment
  • exe.dev VM deployment
  • DigitalOcean droplet
  • Oracle Cloud (Always Free tier)
  • Raspberry Pi (always-on Gateway)
  • macOS VMs (Lume)
  • VPS hosting overview

14-cli.md — CLI Reference (47 pages)

Read this when the user asks about:

  • Any openclaw CLI command
  • CLI overview, global flags, command tree
  • Commands: agent, agents, acp, approvals, backup
  • Commands: browser, channels, config, configure, cron
  • Commands: daemon, dashboard, devices, directory, dns, docs, doctor
  • Commands: gateway, health, hooks, logs, memory, message, models
  • Commands: nodes, onboard, pairing, plugins, qr, reset
  • Commands: sandbox, secrets, security, sessions, setup, skills
  • Commands: status, system, tui, uninstall, update, voicecall, webhooks
  • Shell completion

15-web-uis.md — Web UIs (4 pages)

Read this when the user asks about:

  • WebChat (Gateway WebSocket UI)
  • Dashboard (Control UI)
  • TUI (Terminal UI)

16-security.md — Security (2 pages)

Read this when the user asks about:

  • Threat model (MITRE ATLAS framework)
  • Contributing to OpenClaw security
  • Security threat analysis

17-templates.md — Templates & Workspace Files (8 pages)

Read this when the user asks about:

  • AGENTS.md template (workspace config)
  • IDENTITY.md template (agent identity)
  • SOUL.md template (agent personality)
  • USER.md template (user info)
  • TOOLS.md template (tool notes)
  • HEARTBEAT.md template
  • BOOT.md template (startup instructions)
  • BOOTSTRAP.md template (first-run conversation)

18-reference.md — Reference (15 pages)

Read this when the user asks about:

  • Default AGENTS.md configuration
  • Onboarding wizard reference
  • Token use and costs
  • Prompt caching configuration
  • Transcript hygiene (provider fixups)
  • Session management deep dive
  • SecretRef credential surface
  • RPC adapters (signal-cli, imsg)
  • Device model database
  • API usage and costs
  • Tests and CI pipeline
  • Release checklist
  • Pi development workflow

19-troubleshooting.md — Help & Troubleshooting (8 pages)

Read this when the user asks about:

  • FAQ (comprehensive)
  • General troubleshooting
  • Testing (Vitest suites, Docker runners)
  • Debugging
  • Environment variables
  • Diagnostics flags
  • Helper scripts
  • Node + tsx crash issues

20-experiments.md — Experiments & Design Docs (10 pages)

Read this when the user asks about:

  • Internal plans and proposals
  • PTY and process supervision plan
  • Model config exploration
  • Workspace memory research
  • Session binding plan
  • ACP thread-bound agents plan
  • ACP unified streaming refactor
  • Browser evaluate CDP refactor
  • OpenResponses gateway plan
  • Onboarding config protocol
  • Kilo gateway integration design

Important Rules

  1. NEVER paraphrase code blocks or configuration examples — use them exactly as they appear in the docs
  2. NEVER invent configuration keys or CLI flags — only reference what exists in the documentation
  3. When showing config examples, include the full context (surrounding keys, comments) from the docs
  4. If the docs mention a version requirement or caveat, always include it in your answer
  5. If you cannot find the answer in the documentation files, say so explicitly — do not guess
  6. Always mention which documentation section the answer comes from
  7. For CLI commands, always show the exact syntax from the docs
  8. Each docs/ file has <!-- SOURCE: url --> comments indicating the original page URL — you can reference these for the user

适合场景

01

调用多模型

02

代码和文本生成

03

Agent 推理流程

04

OpenRouter 模型接入

能力概览

能力 1

统一调用多种 LLM

能力 2

支持 Claude、Gemini、Kimi 等模型

能力 3

适合聊天、代码和推理任务

能力 4

可作为 Agent 模型调用入口

平台分布

OpenClaw

91.77%
按下载量换算3,184

安全审计

VirusTotal

通过

ClawScan

可疑

Static analysis

可疑

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills