Token导航 LogoToken导航

Agent Skill · 研究检索

aident-skill(辅助技能)

aident-skill 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 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。

aident-skill 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 64,068。

总安装 306周安装 13GitHub Stars 1下载 107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
unknown

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:aident-skill(辅助技能)
来源仓库:https://github.com/aident-ai/aident-skill
仓库路径:skills/aident-skill
安装命令:
npx skills add https://github.com/aident-ai/aident-skill --skill aident-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 npx skills 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。

skills.shnpx skills
npx skills add https://github.com/aident-ai/aident-skill --skill aident-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。

aident-skill 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 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。

提供 1000+ 集成技能支持,包括 Gmail、Slack、GitHub、Firecrawl、Exa 等,可创建自动化剧本和管理执行。通过 npx skills add 命令从 GitHub 仓库安装,支持自然语言生成剧本和监控运行状态。使用前请检查权限范围、维护状态,注意是否涉及联网、命令执行或文件读写操作。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/aident-skill

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 Codex、Claude、Cursor、Gemini CLI;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 npx skills add https://github.com/aident-ai/aident-skill --skill aident-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

aident-skill 怎么安装?

在终端执行 npx skills add https://github.com/aident-ai/aident-skill --skill aident-skill。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

aident-skill 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。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。

aident-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。

aident-skill 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

aident-skill 用的人多吗?

公开数据是累计安装 306 次,近一周安装 13 次,下载 107 次,GitHub 1 星。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。

aident-skill 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

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。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 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 Codex,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。
  • 需要明确授权才能用到公司项目里的场景。
  • 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
aident-skill

辅助技能 · 本页

3061接触敏感数据未明确未通过

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

Aident (Dual-mode: MCP preferred, HTTPS fallback)

Access 1k+ integrations (Gmail, Slack, GitHub, Firecrawl, Exa, etc.), create and manage automation playbooks, discover templates, and monitor all running automations from a command center dashboard.

What this skill does

  • Search and execute 1000+ integration skills (email, messaging, project management, web scraping, etc.)
  • Generate, execute, and manage automation playbooks from natural language
  • Browse and instantiate pre-built playbook templates
  • Monitor active automations and track execution results
  • Connect new third-party integrations on-the-fly

It supports two execution modes:

  1. MCP mode (preferred): use MCP tools from the aident server.
  2. REST API fallback mode: call POST /api/mcp/rest with {tool, arguments} when MCP tools are unavailable.

Decide which mode to use

Use MCP mode if the client has MCP tools available named like:

  • capability_search, skill_list, skill_get_info, skill_execute
  • playbook_list, playbook_generate, playbook_execute
  • template_search, template_list, template_instantiate
  • integration_status, integration_connect
  • dashboard_active_playbooks, execution_list

Otherwise use REST API fallback.

MCP mode (preferred)

When: The client is connected to the Aident MCP server and can call tools like capability_search, playbook_list, etc.

Setup: See references/mcp.md for client configuration.

Workflow:

  1. Collect required inputs from the user.
  2. Call the relevant tool(s) directly.
  3. Handle errors:

- If auth error: call auth_logout, delete ~/.aident/credentials.json, then reconnect via OOB flow. - If missing integration: call integration_connect to connect it, then retry.

  1. Return results in a clean format.

REST API fallback mode

When: No Aident MCP tools are available in the client.

Credentials file

Authentication is persisted in ~/.aident/credentials.json so tokens survive across sessions:

{
  "base_url": "https://app.aident.ai",
  "client_id": "<oauth_client_id>",
  "access_token": "<bearer_token>",
  "refresh_token": "<refresh_token>",
  "expires_at": "<ISO8601_timestamp>"
}
Step 1: Load credentials
  1. Check if AIDENT_TOKEN env var is set. If yes, use it directly as the Bearer token and skip to Step 3. (This is an advanced override -- do not ask the user for it.)
  2. Read ~/.aident/credentials.json. If the file exists and has a non-empty access_token:

- If expires_at is in the past and refresh_token is present, go to Step 2b (refresh). - Otherwise skip to Step 3.

  1. If the file does not exist or has no access_token, go to Step 2a (first-time setup).
Step 2a: First-time authentication (OOB flow)

Run these steps automatically -- never ask the user to provide a token manually.

  1. Resolve the base URL: use AIDENT_BASE_URL env if set, otherwise https://app.aident.ai.
  2. Create the credentials directory if it does not exist: mkdir -p ~/.aident
  3. Register an OAuth client: curl -s -X POST $BASE_URL/api/mcp/oauth/register \ -H "Content-Type: application/json" \ -d '{"redirect_uris": ["'"$BASE_URL"'/mcp/oob"], "client_name": "aident-skill-cli", "grant_types": ["authorization_code", "refresh_token"], "response_types": ["code"], "token_endpoint_auth_method": "none"}' Save client_id from the JSON response.
  4. Open the authorization URL in the user's browser: $BASE_URL/api/mcp/oauth/authorize?response_type=code&client_id=$CLIENT_ID&redirect_uri=$BASE_URL/mcp/oob Use open (macOS), xdg-open (Linux), or start (Windows) to launch the browser.
  5. Tell the user: "I've opened Aident in your browser. Please log in and click Approve, then paste the access token shown on screen back here."
  6. When the user pastes the token, write ~/.aident/credentials.json: {"base_url": "$BASE_URL", "client_id": "$CLIENT_ID", "access_token": "<pasted_token>", "refresh_token": "", "expires_at": ""}
  7. Proceed to Step 3.
Step 2b: Refresh an expired token
curl -s -X POST $BASE_URL/api/mcp/oauth/token \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "grant_type=refresh_token&client_id=$CLIENT_ID&refresh_token=$REFRESH_TOKEN"

If refresh succeeds, update access_token, refresh_token, and expires_at in ~/.aident/credentials.json. If refresh fails (e.g. token revoked or expired), delete the credentials file and go back to Step 2a.

Step 3: Call tools

Use the access_token from the credentials file (or AIDENT_TOKEN env) and base_url (or AIDENT_BASE_URL env, default https://app.aident.ai):

curl -s -X POST $BASE_URL/api/mcp/rest \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $ACCESS_TOKEN" \
  -d '{ "tool": "<tool_name>", "arguments": { ... } }'

Parse the result field from the JSON response.

On HTTP 401: The token has expired. Go to Step 2b to refresh, then retry the request. If refresh also fails, go to Step 2a.

Available Tools (22)

Auth (2)

  • auth_status -- Check authentication status (always accessible)
  • auth_logout -- Revoke access token and log out. After calling this, delete ~/.aident/credentials.json so the next request triggers a fresh OOB flow.

Skills & Discovery (4)

  • capability_search -- Search skills and integrations by query, type, or scope using hybrid search
  • skill_list -- List available skills with pagination
  • skill_get_info -- Get detailed metadata including input/output schemas and required integrations
  • skill_execute -- Execute a skill with validated input; prompts for missing integrations

Integrations (2)

  • integration_status -- Check which integrations are connected
  • integration_connect -- Initiate connection to a third-party service via OAuth

Playbooks (6)

  • playbook_list -- List your playbooks with status and trigger info
  • playbook_get_info -- Get playbook details including content and trigger configuration
  • playbook_generate -- Generate a new playbook from a natural language description
  • playbook_execute -- Execute a playbook (returns execution ID for tracking), or send a follow-up message to an existing execution via executionSessionId
  • playbook_update_trigger -- Enable or disable playbook triggers
  • playbook_execution_history -- Get execution history for a specific playbook

Templates (4)

  • template_search -- Search for playbook templates by keyword or category
  • template_list -- List available templates with optional category filtering
  • template_get_info -- Get detailed template information
  • template_instantiate -- Create a new playbook from a template

Dashboard (4)

  • dashboard_active_playbooks -- List playbooks with active triggers or running executions
  • execution_get_details -- Get execution details including status and messages
  • execution_list -- List recent executions across all playbooks
  • execution_get_messages -- Get simplified chat messages for progress polling

Safety & privacy

  • Never request secrets in plain text if the platform has secret storage.
  • If the user pastes a token, suggest they rotate it and store it securely.
  • Only send necessary fields to the service.
  • All tokens are scoped -- request only the permissions you need.

Examples

Example 1: MCP mode

User: "Find skills for sending emails and send a meeting summary to team@example.com"

Assistant workflow:

  1. Call capability_search with {"query": "send email"}
  2. Review results, pick best match (e.g. gmail_send_email)
  3. Call skill_execute with the skill and input
  4. Present confirmation to user

Example 2: REST fallback, first-time user

User: "Send an email to team@example.com with today's meeting notes"

Assistant workflow (no ~/.aident/credentials.json found):

  1. Register OAuth client with the Aident server
  2. Open browser to authorization page
  3. Tell user: "I've opened Aident in your browser. Please log in and click Approve, then paste the access token shown on screen back here."
  4. User pastes token
  5. Save credentials to ~/.aident/credentials.json
  6. Call capability_search for email skills via REST
  7. Call skill_execute to send the email via REST
  8. Confirm to user

Example 3: REST fallback, returning user

User: "List my playbooks"

Assistant workflow (reads existing ~/.aident/credentials.json):

  1. Load access_token and base_url from credentials file
  2. POST to $BASE_URL/api/mcp/rest with {"tool": "playbook_list", "arguments": {}}
  3. Parse the result field from response
  4. Present playbook list to user

Security

  • OAuth 2.1 + PKCE: Industry-standard authentication with automatic token refresh
  • Scoped Access: Category-based permissions (skills, integrations, playbooks, templates, dashboard)
  • Revocable: Revoke access anytime from Settings
  • Integration-Aware: Missing integrations prompt for connection rather than failing silently

Documentation

Support

License

MIT License - See LICENSE file for details

平台分布

Codex

34.03%
按下载量换算36

Claude

29.73%
按下载量换算32

Cursor

20.63%
按下载量换算22

Gemini CLI

10.24%
按下载量换算11

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

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。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