RPA Caller
通过 HTTP 请求调用 RPA 自动化功能,让机器人完成指定任务。 当用户提到「RPA」「自动化」「机器人执行」「让机器人帮我」「自动填表」「自动点击」 「批量处理」「自动化流程」「运行任务」「触发流程」等词时,必须使用此 skill。 适用场景:触发 RPA 任务、查询任务状态、停止任务、传参数给 RPA 流程。
MIT-0 · Free to use, modify, and redistribute. No attribution required.
⭐ 0 · 36 · 1 current installs · 1 all-time installs
MIT-0
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description match the implementation: the SKILL instructs the agent to map user intents to HTTP endpoints and call an RPA service. The included API map defines relevant endpoints and parameters; nothing requested (no unrelated cloud creds or weird binaries) is obviously out of scope.
Instruction Scope
Instructions let the agent collect sensitive values (BASE_URL, API_KEY) from the user and 'remember' them in conversation; they also allow constructing and issuing arbitrary HTTP requests to any BASE_URL. This is expected for an RPA integrator, but it enables the agent to contact internal services or exfiltrate data if misused. The SKILL also references an Excel file path (references/rpa-api-map.xlsx) while the bundle contains references/rpa-api-map.json — a minor inconsistency that could cause runtime confusion.
Install Mechanism
Instruction-only skill with no install step and no code files to write or execute. That lowers the risk from the install mechanism side.
Credentials
SKILL.md expects BASE_URL and API_KEY to be provided, but the registry metadata lists no required env vars or primary credential. Asking users to supply an API_KEY is normal for authenticated RPA APIs, but the skill does not declare or restrict where/how that secret is stored (it instructs to 'remember' it in conversation), which is a privacy/operational concern.
Persistence & Privilege
always is false and disable-model-invocation is false (normal). The skill instructs the agent to keep settings in the dialogue context, but it does not request permanent system-level privileges or modify other skills.
What to consider before installing
This skill is plausible for calling an RPA service, but review these points before enabling it: (1) The skill asks you to provide BASE_URL and an API_KEY — prefer a least-privilege/scoped token and never paste high-privilege secrets into long-lived public chat logs. (2) The agent will construct and send HTTP requests to whatever BASE_URL you give it; do not point it at sensitive internal services unless you trust the agent and network isolation (this can be used to access internal endpoints). (3) The SKILL.md mentions an Excel file (rpa-api-map.xlsx) but the package contains a JSON map — verify the API map is correct and that endpoints are what you expect. (4) If you worry about automated calls, restrict the agent's ability to act autonomously (require explicit user confirmation for each run) and prefer manual paste of short-lived tokens. (5) Audit the RPA endpoints and logs on the server side, and consider network-level controls (firewall, allowlist) to limit where the agent can connect.Like a lobster shell, security has layers — review code before you run it.
Current versionv1.0.0
Download ziplatest
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
SKILL.md
RPA HTTP 调用 Skill
本 Skill 让 Claude 能够通过 HTTP 请求调用 RPA 系统,完成各类自动化任务。
RPA 功能清单和对应 API 定义在 references/rpa-api-map.xlsx 中,使用前必须先读取。
核心工作流程
Step 1:理解用户意图
明确用户想让 RPA 完成什么任务,例如:
- "帮我自动填写报销单"
- "批量下载这些文件"
- "打开系统并录入数据"
Step 2:查找对应 RPA 功能
从 Excel 功能清单(references/rpa-api-map.xlsx)中查找:
- 匹配用户意图的功能名称
- 对应的 HTTP Method 和 Endpoint
- 必填参数 / 可选参数
- 预期返回结果
如果找不到匹配功能,告知用户并列出所有可用功能供其选择。
Step 3:收集必填参数
根据功能清单,向用户确认所有必填参数。例如:
- 目标文件路径
- 要填写的数据内容
- 时间范围
- 目标系统账号(如需要)
对于敏感参数(密码、token),提示用户注意安全,避免明文存储。
Step 4:构建 HTTP 请求
Method: [GET / POST / PUT / DELETE]
URL: {BASE_URL}{endpoint}
Headers:
Content-Type: application/json
Authorization: Bearer {API_KEY} ← 如功能清单中有鉴权要求
Body (JSON):
{
"task_id": "唯一任务ID(可用时间戳)",
"params": {
// 根据功能清单填入具体参数
}
}
Step 5:展示请求并确认
在执行前,以结构化方式展示完整的请求内容,请用户确认后再发送。
示例展示格式:
📋 即将调用 RPA 功能:【批量下载文件】
接口:POST http://localhost:8088/api/rpa/download-files
参数:
- source_dir: /data/reports/2024/
- file_pattern: *.pdf
- dest_dir: /output/
是否确认执行?
Step 6:发起请求并处理响应
发起请求后,根据响应状态给出反馈:
| 响应状态 | 处理方式 |
|---|---|
| 200 / 任务提交成功 | 告知任务ID,提示如何查询状态 |
| 202 Accepted | 异步任务已接受,等待执行 |
| 400 参数错误 | 列出错误字段,引导用户修正 |
| 401 / 403 鉴权失败 | 提示检查 API_KEY 或权限 |
| 404 接口不存在 | 检查 BASE_URL 和 endpoint 是否正确 |
| 500 服务端错误 | 建议检查 RPA 服务是否正常运行 |
| 超时 / 无响应 | 建议检查 RPA 服务的连接状态 |
Step 7:任务状态追踪(如支持)
如果功能清单中有状态查询接口,主动提示用户可以查询执行进度:
任务已提交!Task ID: rpa_20240315_143022
可以告诉我「查询任务状态」来追踪执行进度 🤖
全局配置(首次使用时设置)
使用本 Skill 前,需要确认以下配置信息:
BASE_URL: http://your-rpa-server:8088 ← RPA 服务地址
API_KEY: your-api-key-here ← 鉴权 Token(如有)
如果用户没有提供,主动询问这两个值,并在对话中记住它们。
常用功能速查
以下为示例,实际功能以 Excel 功能清单为准
| 用户说 | 对应功能 |
|---|---|
| 帮我填表 / 录入数据 | data-entry |
| 下载文件 / 批量下载 | file-download |
| 截图 / 截取屏幕 | screenshot |
| 打开网页 / 访问系统 | open-browser |
| 查询任务进度 | task-status |
| 停止任务 | task-stop |
注意事项
- 安全:不要在对话中明文显示完整的密码或高权限 Token
- 幂等性:每次请求使用唯一 task_id(建议用时间戳)避免重复执行
- 异步任务:大多数 RPA 任务是异步的,提交后需要轮询状态接口
- 错误重试:网络错误可以重试,但业务逻辑错误需要修正参数后再试
- 日志:建议用户在 RPA 系统侧开启日志,便于排查问题
参考文件
references/rpa-api-map.xlsx— RPA 功能清单,包含所有接口定义和参数说明- 如需新增功能,按 Excel 模版格式填写后告知 Claude 重新加载
Files
3 totalSelect a file
Select a file to preview.
Comments
Loading comments…
