Token导航 LogoToken导航

Agent Skill · 研究检索

copilot-studio-agent(GitHub Copilot studio Agent 搜索)

copilot-studio-agent 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。

copilot-studio-agent 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为调用外部服务;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 14,809。

总安装 8,288周安装 356GitHub Stars 公开资料未说明下载 2,905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:copilot-studio-agent(GitHub Copilot studio Agent 搜索)
来源仓库:https://github.com/saikiranparamkusham09/copilot-studio-agent
安装命令:
openclaw skills install copilot-studio-agent
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 copilot-studio-agent

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。

通过引导式 Web 应用程序流程在 Microsoft Copilot Studio 中创建和配置自定义代理,包括触发器、知识、工具、主题和发布。

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 copilot-studio-agent;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

    它需要调用外部服务,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

copilot-studio-agent 怎么安装?

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

copilot-studio-agent 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。

copilot-studio-agent 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

copilot-studio-agent 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。

copilot-studio-agent 用的人多吗?

公开数据是累计安装 8,288 次,近一周安装 356 次,下载 2,905 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。

copilot-studio-agent 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

  • 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

  • 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
copilot-studio-agent

GitHub Copilot studio Agent 搜索 · 本页

8,288—调用外部服务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

Skill: Create Copilot Studio Agents from Scratch

Create agents in Microsoft Copilot Studio using the web app. This skill guides you through the complete agent creation process based on Microsoft Learn documentation.

Quick Reference

Documentation: https://learn.microsoft.com/en-us/microsoft-copilot-studio/authoring-first-bot

Key Limits:

  • Agent name: max 42 characters (no angle brackets < >)
  • Description: max 1,024 characters
  • Instructions: max 8,000 characters
  • Icon: PNG format, <72KB, max 192×192 pixels

Steps to Create an Agent

1. Start at Copilot Studio

  1. Sign in to https://copilotstudio.microsoft.com/
  2. Switch to the desired environment (if needed)
  3. Optionally configure: primary language, solution, schema name

2. Create the Agent (3 Options)

Option A - Natural Language:

  • On Home or Agents page, enter a brief description of what you want the agent to do (up to 1,024 chars)
  • AI generates name, description, instructions, triggers, channels, knowledge sources, tools

Option B - Create from Scratch:

  • Home page: Select Create an agent under "Start building from scratch"
  • OR Agents page: Select Create blank agent

Option C - Advanced Create:

  • Agents page: Select dropdown next to Create blank agent → Advanced create
  • Configure primary language, solution, and schema name before creating

3. Configure Basic Features

After provisioning, configure:

FeatureDescription
Primary AI ModelChange agent model
TriggersDefine what activates the agent (About triggers)
Knowledge SourcesAdd enterprise data, websites, SharePoint, Dataverse (Knowledge)
ToolsAdd Power Automate flows, connectors (Add tools)
Other AgentsAdd sub-agents (Add agents)
TopicsDefine conversation flows (Create topics)
Starter PromptsAdd suggested prompts for users

4. Edit Agent Basics

Rename & Edit Description:

  1. Go to Overview page → Details section
  2. Select Edit → enter name/description → Save

Edit Instructions:

  1. Overview page → Instructions section
  2. Select Edit → refine instructions → Save

Change Icon:

  1. Select agent icon in top bar → Change icon
  2. Choose PNG image → Save

Change Language/Solution:

  1. Select gear icon in description box
  2. Under Agent settings: select primary language
  3. Advanced settings: choose solution and schema name
  4. Update

⚠️ Note: Cannot change primary language after creation. Can change region or add secondary languages.

5. Test & Publish

  1. Test your agent
  2. Improve based on testing
  3. Publish to channels

Topic Types

Node TypePurpose
MessageSend a message to user
QuestionAsk user for input
Adaptive CardInteractive cards with buttons/inputs
ConditionBranch conversation based on logic
VariableSet, parse, or clear variables
Topic ManagementRedirect, transfer, or end conversation
ToolCall Power Automate flows, connectors
AdvancedGenerative answers, HTTP requests, events

Topic Creation (Web App)

  1. Go to Topics page → close test panel
  2. Add a topic → From blank
  3. Add trigger phrases (5-10 recommended)

- Use Add icon or upload a file - Each phrase on separate line

  1. Select Details → add name, display name, description
  2. Save

Knowledge Sources

SourceDescription
Public WebsiteBing search on provided URLs (max 25 in generative mode)
DocumentsUploaded to Dataverse
SharePointConnect via URL
DataverseQuery tables in environment
Enterprise DataMicrosoft Search via connectors

Delete an Agent

Web App:

  1. Go to Agents page
  2. Select agent → … menu → Delete
  3. Enter agent name to confirm

Authentication

  • Agents automatically use Microsoft Entra ID (Azure AD) authentication
  • Can configure SSO so users don't need manual sign-in
  • See: Configure SSO

Event Triggers (Advanced)

Event triggers allow agents to act autonomously in response to external events—without user input.

Requirements

How Event Triggers Work

  1. Event occurs → External system (SharePoint, Planner, etc.) triggers an event
  2. Trigger sends payload → JSON/message containing event info + instructions
  3. Agent executes → Agent reads payload and calls appropriate actions/topics

Available Event Triggers

TriggerSourceExample
When a row is added/modified/deletedDataverseNew record in table
When a file is createdOneDriveNew file uploaded
When an item is createdSharePointNew list item
When a task is completedMicrosoft PlannerTask marked complete
RecurrenceScheduledTime-based trigger (every X minutes)

Add an Event Trigger

  1. Go to Overview → Triggers section
  2. Select Add trigger
  3. Choose the desired trigger
  4. Authenticate (uses agent maker's credentials)
  5. Configure parameters and define trigger payload
  6. Define actions/topics for the agent to call in response

Trigger Payload

The payload is a JSON/plain text message sent to the agent containing:

  • Event information (data from the source)
  • Instructions on how to act

Default payload example: *"Use content from Body"*

Custom payload example: *"Summarize the changes and send to the user"*

Agent Instructions vs Payload Instructions

ApproachUse Case
Agent InstructionsGeneral behavior, simple agents with few triggers
Payload InstructionsComplex agents with multiple triggers/goals

Test a Trigger

  1. Run the triggering event (e.g., create a Planner task)
  2. On Overview page, select Test trigger icon beside the trigger
  3. Choose the instance → Start testing
  4. Use Activity Map to see agent's reaction

⚠️ Before publishing: Agent won't react automatically until published.

Modify a Trigger

  1. Overview → locate trigger → … → Edit in Power Automate
  2. Modify parameters, payload content, variables
  3. Save changes in Power Automate

Data Protection & Security

  • Event triggers use agent maker's credentials only
  • Users may access data using author's authorization
  • Review data protection best practices before publishing

Billing Note

Each trigger activation counts as a message toward Copilot Credits. A recurrence trigger running every 10 minutes = ~4,320 messages/month.


Power Automate Integration (Tools)

Agents can call Power Automate cloud flows as tools to perform actions and get data.

Requirements for Agent Flows

RequirementDetails
TriggerMust use When an agent calls the flow
ResponseMust include Respond to the agent action
ModeReal-time (Async toggle = Off)
TimeoutMust respond within 100 seconds
Flow Run LimitUp to 30 days (after response)

Create an Agent Flow

Option 1: From Topic

  1. Go to Topics → open your topic
  2. Select Add node → Add a tool
  3. Basic tools tab → New Agent flow
  4. Template opens with required trigger + response action
  5. Publish → Go back to agent
  6. Action node added to topic → Save

Option 2: From Overview

  1. Overview → Tools section → Add tool
  2. Select Agent flow → create new or select existing

Configure Flow Inputs

In the When an agent calls the flow trigger, add input parameters:

ParameterTypeExample
CityText"Seattle"
ZIP codeNumber98101
DateDateTime"2024-01-15"

Configure Flow Outputs

In Respond to the agent action, define output parameters:

Output ParameterTypeVariable Name
day_summaryTextDay Summary
locationTextLocation
chance_of_rainNumberDay Rain Chance

Use Flow in Topic

  1. In topic authoring canvas, add Tool node
  2. Select the flow
  3. Map inputs (from user questions or variables)
  4. Use outputs in subsequent nodes

Modify Existing Flow for Agent Use

  1. Open flow in Power Automate
  2. Replace trigger with When an agent calls the flow
  3. Add Respond to the agent action
  4. Configure inputs/outputs as needed
  5. Save and test

Speed Up Flow Execution

  • Use Express Mode for faster agent flow execution
  • Optimize queries and logic
  • Keep typical runs under 100 seconds
  • Move long-running actions after Respond to Copilot action

Connection Management

Auth TypeDescription
Maker credentialsDefault - uses author's account
User credentialsUsers run with their own permissions (generative orchestration)

For user credentials: Configure flow's "Run-only permissions" to Provided by run-only user.

⚠️ CMK Environments: Cloud flows can't run with customer credentials. Use specific connections instead.


Publishing with Triggers

Pre-Publish Warning

When publishing an agent with event triggers, you'll see a warning about author credentials. Users may access data using the agent maker's authorization.

After Publishing

  • Agent reacts automatically when triggers activate
  • Monitor via Activity page
  • Each trigger event creates a log entry

Test Mode vs Published

ModeTrigger Behavior
TestManual activation required
PublishedAutomatic activation on events

Related Docs

Related Docs

平台分布

OpenClaw

79.55%
按下载量换算2,311

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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