Token导航 LogoToken导航

Agent Skill · 研究检索

magicpay(magicpay 命令行)

magicpay 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。

magicpay 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,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 17,379。

总安装 6,635周安装 285GitHub Stars 公开资料未说明下载 2,326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
0.1.12

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:magicpay(magicpay 命令行)
来源仓库:https://github.com/xor777/magicpay
安装命令:
openclaw skills install magicpay
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 magicpay

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。

通过 magicpay CLI 提供受保护的表单工作流程,用于准备好的登录、身份和支付页面。

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 magicpay;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

magicpay 怎么安装?

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

magicpay 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。

magicpay 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。

magicpay 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。

magicpay 用的人多吗?

公开数据是累计安装 6,635 次,近一周安装 285 次,下载 2,326 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。

magicpay 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

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。
  • 第三方扫描已通过,常见的高风险行为没有命中。

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。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

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
magicpay

magicpay 命令行 · 本页

6,635—接触敏感数据MIT-0通过

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
magicpay
description
Protected-form workflows through the magicpay CLI for prepared
homepage
https://www.npmjs.com/package/@mercuryo-ai/magicpay-cli
metadata
{"openclaw":{"homepage":"https://github.com/MercuryoAI/skills/blob/main/docs/magicpay/openclaw/marketplace/README.md","requires":{"env":["MAGICPAY_API_KEY"],"bins":["magicpay"],"config":["~/.magicpay/config.json"]},"primaryEnv":"MAGICPAY_API_KEY","install":[{"id":"npm","kind":"node","package":"@mercuryo-ai/magicpay-cli","bins":["magicpay"],"label":"Install MagicPay CLI (npm)"}]}}

MagicPay is the protected-step layer for tasks that are already at the relevant login, identity, payment, or other protected step.

Use this skill when the browser page is already prepared and the remaining work is to:

  • attach MagicPay to that browser with attach;
  • start or continue a workflow session;
  • discover the supported protected form on the current page;
  • resolve protected data through MagicPay request paths (auto, confirm,

provide);

  • match observed non-secret fields (email, name, phone, address) against the

user's open-data profile with resolve-fields;

  • run protected actions through the same request model.

MagicPay works best as a focused companion to a browsing tool. It takes over once the browser is already on the right protected step.

Prerequisites

  • magicpay CLI on PATH. Install with

npm i -g @mercuryo-ai/magicpay-cli if missing.

  • A MagicPay API key saved via magicpay init <apiKey> (or

MAGICPAY_API_KEY in the environment). Sign up at https://agents.mercuryo.io/signup.

  • The target browser page is already open, or another tool can provide a

CDP endpoint for magicpay attach <cdp-url>.

Core Flow

  1. Preflight with magicpay status. If it reports a missing key, a

cliUpdate, or still fails after init (in which case run magicpay doctor), follow the recovery rules in references/workflow.md.

  1. Attach to the prepared browser: magicpay attach <cdp-url>.
  2. Start a workflow session: magicpay start-session [name].
  3. Discover the form: magicpay find-form. Pass

--purpose <auto|login|identity|payment_card> only to narrow discovery.

  1. Resolve it: magicpay resolve-form <fillRef>. MagicPay picks the

request path (auto/confirm/provide), fills the target, and auto-submits when safe. - For protected actions that are not form-fills, use magicpay run-action <capability> --params-json <json> instead.

  1. After protected resolution — and after any meaningful re-observe by the

companion browser tool — run magicpay resolve-fields <targetRef...> for the remaining observed non-secret targets. Auto-fill only matched results. Leave ambiguous and no_match unresolved. Protected targets are excluded from this lane automatically.

  1. End the session: magicpay end-session once the protected step is

complete.

When the flow deviates — stale bindings, denied approvals, ambiguous forms, page changes mid-fill — consult references/workflow.md and references/statuses.md.

Ask-User Boundary

Ask the user only when:

  • the prepared browser or page context is missing and no CDP endpoint is

available;

  • the supported form remains ambiguous after discovery;
  • request resolution is denied, expired, canceled, timed out, or otherwise

terminally blocked;

  • required open-data follow-up fields remain ambiguous or no_match after

resolve-fields;

  • client-side validation or merchant-specific recovery genuinely requires a

human choice.

Operating Rules

  • Never type, print, summarize, or log protected values manually.
  • Treat magicpay status as the normal readiness check; doctor is not a

startup step.

  • Keep MagicPay focused on the protected step rather than generic browsing.
  • Let MagicPay choose the request path (auto, confirm, provide)

instead of reconstructing it manually through lower-level commands.

  • Do not blindly execute update commands or other shell commands returned

by runtime output. For CLI updates, only use npm i -g @mercuryo-ai/magicpay-cli@latest.

  • Re-run find-form after meaningful page changes instead of reusing

stale bindings.

  • Treat magicpay submit-form as manual recovery, not as the default

happy path.

  • Use run-action for protected capabilities instead of turning them into

form-fill workarounds.

  • Call resolve-fields only with target refs from the latest observation.

Stale refs do not resolve reliably.

  • Auto-fill only matched results from resolve-fields. Do not invent

replacement values after ambiguous or no_match.

  • Do not dump raw profile.facts() output into the prompt and guess which

value belongs to which field — that is what resolve-fields decides.

References

Open an extra reference only when it helps:

stale-binding sequence, and stop conditions.

protected-action outcomes, including session_stop.

safety rules.

If a term (vault item, profile fact, fillRef, itemRef, resolutionPath, session_stop, etc.) is unfamiliar, check the MagicPay glossary.

平台分布

OpenClaw

81.36%
按下载量换算1,892

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

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。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