← Back to Skills Marketplace
jeandoom

Link All

by Rainco · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
84
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install link-all
Description
帮助用户调研、选择并配置外部平台链接方案,支持插件、CLI或API方式,完成认证和首次调用测试。
README (SKILL.md)

link-all - 外部平台链接与管理

概述

这是一个专门帮助用户链接、访问、调用、管理外部平台的 skill。

职责

  1. 寻找合适的工具和方案 - 完成用户指定平台的链接、授权、访问、管理
  2. 平台调研 - 了解平台背景信息,查找相关技术文档(API/CLI)
  3. 方案适配 - 优先查找 OpenClaw 插件,其次 CLI 工具,最后梳理 API
  4. 用户确认 - 将找到的方案发给用户选择
  5. 脚本开发 - 如需 API 则编写相关脚本
  6. 技能转化 - 首次成功链接后询问用户是否创建为 skill

工作流程

步骤 1:了解平台背景

当用户提供要访问的平台时:

  • 搜索该平台的基本信息
  • 了解平台的定位、主要功能、认证方式
  • 确定平台是否支持 API/CLI 访问

步骤 2:查找技术文档

  • 搜索平台官方 API 文档
  • 搜索平台官方 CLI 工具
  • 搜索开发者社区的相关资源

步骤 3:适配方案优先级

优先级 1: OpenClaw Plugins → 优先查找是否已有现成插件
优先级 2: CLI 工具 → 查找官方或社区提供的命令行工具
优先级 3: API 文档 → 梳理 REST API 文档,编写调用脚本

步骤 4:用户确认

将找到的方案整理后发给用户:

  • 方案 1:使用 XXX 插件(如果存在)
  • 方案 2:使用 XXX CLI 工具
  • 方案 3:通过 API 调用(需要开发脚本)

等待用户确认使用哪个方案。

步骤 5:执行链接

  • 如果是插件:帮助安装配置
  • 如果是 CLI:帮助安装并配置认证
  • 如果是 API:编写脚本并完成首次调用测试

步骤 6:技能转化(重要!)

首次成功链接后,必须询问用户:

✅ 平台链接成功!是否需要将此平台的链接方案创建为独立的 skill?
- "要" / "需要" → 帮用户梳理方案并创建 skill
- "不要" / "不需要" → 结束

如果用户需要创建 skill:

  1. 梳理完整的链接方案(认证方式、配置参数、常用操作)
  2. ~/.openclaw/skills/ 下创建新的 skill 目录
  3. 编写 SKILL.md 文件
  4. 告知用户 skill 已创建

输出格式

平台信息报告

## 平台:XXX

### 基本信息
- 官网:xxx
- 类型:API/CLI/SDK
- 认证方式:API Key/OAuth/Token

### 技术文档
- API 文档:xxx
- CLI 工具:xxx
- SDK:xxx

### 推荐方案
1. 方案一:xxx
2. 方案二:xxx
3. 方案三:xxx

链接成功报告

## ✅ XXX 平台链接成功

### 配置信息
- 认证方式:xxx
- 配置参数:xxx
- 测试结果:xxx

### 常用操作
- 操作 1:xxx
- 操作 2:xxx

### 是否需要创建 skill?
请回复 "要" 或 "不要"

示例对话

用户:帮我连接 GitHub

## 平台:GitHub

### 基本信息
- 官网:https://github.com
- 类型:代码托管平台
- 认证方式:Personal Access Token (PAT)

### 技术文档
- API:https://docs.github.com/rest
- CLI:gh (GitHub CLI)

### 推荐方案
1. **方案一(推荐)**:使用 `gh` CLI 工具 - 官方出品,易用
2. **方案二**:使用 GitHub REST API - 灵活,可开发脚本

请选择方案(1 或 2):

用户:选 1

:帮你安装配置 GitHub CLI...

(链接成功后)

## ✅ GitHub 链接成功

- 用户:xxx
- 认证:SSH Key 已配置
- 测试:✓ 已验证

是否需要将此平台的链接方案创建为独立的 skill?
- 回复 "要" 或 "不要"

注意事项

  • 必须按照步骤执行,不能跳过用户确认环节
  • 首次链接成功后必须询问用户是否需要创建 skill
  • 如果用户选择创建 skill,需要完整记录配置和操作步骤
  • 保持文档清晰,方便后续复用
Usage Guidance
This skill appears to do what it says (research and wire up external platforms), but there are important cautions: - The SKILL.md tells the agent to create new skill files under your home (~/.openclaw/skills/). The registry metadata did not declare this config path — expect the agent to write files to your filesystem if you accept creation. - The skill will want to obtain and use platform credentials (API keys, OAuth tokens, SSH keys) and run initial API calls. Ask the skill how it will store any credentials it uses. Prefer that it does NOT persist secrets in plaintext inside created SKILL.md or other files. - Before allowing creation of a new skill, request to review the exact files the agent will write and their contents, and require explicit consent for writing them. Consider requiring the agent to produce instructions and a patch you can apply manually rather than letting it write files directly. - Limit risk by providing minimal-scope or ephemeral credentials for testing, or by performing the sensitive setup steps locally yourself based on the agent's guidance. - If you plan to let the agent run commands autonomously, consider running it in a sandboxed environment or review created files immediately after creation. If you want, I can: (1) produce a checklist of questions to ask the agent before giving it filesystem/credential access, (2) suggest secure templates for SKILL.md that avoid embedding secrets, or (3) draft a consent prompt you can require before any file writes.
Capability Analysis
Type: OpenClaw Skill Name: link-all Version: 1.0.0 The 'link-all' skill is designed to automate the integration of external platforms by researching APIs/CLIs, installing software, writing scripts, and creating new OpenClaw skills in `~/.openclaw/skills/`. While the instructions in `SKILL.md` emphasize user confirmation, the skill possesses high-risk capabilities including arbitrary code execution (via CLI installation and script development) and persistence (via skill creation). These behaviors are inherent to its stated purpose but represent a significant attack surface if the agent is directed toward malicious resources during its research phase.
Capability Assessment
Purpose & Capability
The skill's name/description (finding and wiring external platform access via plugin/CLI/API) matches the runtime instructions (research, present options, implement chosen approach). However the SKILL.md explicitly instructs creating new skill directories under ~/.openclaw/skills/ (persistence/config write) while the registry metadata lists no required config paths — that discrepancy is an incoherence to be aware of.
Instruction Scope
Instructions direct the agent to research platforms, install/configure CLIs or plugins, write API-calling scripts, and perform first-call tests. They also mandate asking the user whether to 'create a skill' and, if requested, to write files in ~/.openclaw/skills/ including a SKILL.md. The doc gives no guidance on how credentials should be obtained, used, or stored securely, and does not restrict what the created SKILL.md may contain (risk of persisting secrets or sensitive configuration).
Install Mechanism
No install spec and no code files — this is instruction-only, which is the lowest install risk. The skill does not pull code or binaries itself.
Credentials
The registry lists no required environment variables or config paths, but runtime behavior implies obtaining and using platform credentials (API keys, OAuth tokens, SSH keys) and possibly writing configuration into ~/.openclaw/skills/. Requesting or persisting secrets is reasonable for the task, but the absence of declared config paths and lack of secure-handling guidance is disproportionate and ambiguous.
Persistence & Privilege
The skill explicitly instructs writing new skill directories and SKILL.md files into ~/.openclaw/skills/, which modifies the agent's skill set/persistence. Although creating skills is within the realm of its purpose, this action is not declared in metadata and, combined with the agent's ability to run API tests with credentials, increases risk (new skill code could be introduced to the agent environment).
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install link-all
  3. After installation, invoke the skill by name or use /link-all
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of link-all: a skill for connecting and managing external platforms. - Guides users to link, access, authorize, and manage external platforms. - Researches target platforms and collects official documentation for APIs, CLI tools, or plugins. - Suggests linking solutions in order: OpenClaw plugin > CLI tool > API scripting. - Sends found options to users for confirmation before executing. - Assists with configuration, scripting, and first-time connection testing. - After successful linking, prompts users to optionally save the solution as a new skill for reuse. - Provides standardized output templates for reports and conversations.
Metadata
Slug link-all
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Link All?

帮助用户调研、选择并配置外部平台链接方案,支持插件、CLI或API方式,完成认证和首次调用测试。 It is an AI Agent Skill for Claude Code / OpenClaw, with 84 downloads so far.

How do I install Link All?

Run "/install link-all" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Link All free?

Yes, Link All is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Link All support?

Link All is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Link All?

It is built and maintained by Rainco (@jeandoom); the current version is v1.0.0.

💬 Comments