Command Palette

Search for a command to run...

UnylyUnyly
Весь каталог

Wais

БесплатноНе проверен

Universal MCP server for WAIS (Web Agent Interaction Standard) that provides 5 generic tools to discover, register, execute actions, confirm high-risk actions,

GitHubEmbed

Описание

Universal MCP server for WAIS (Web Agent Interaction Standard) that provides 5 generic tools to discover, register, execute actions, confirm high-risk actions, and check status on any WAIS-compatible site.

README

Universal MCP server for WAIS (Web Agent Interaction Standard) — 5 generic tools that work with any WAIS-compatible site.

Works with Claude, ChatGPT, Gemini, Cursor, Windsurf, VS Code Copilot, and any MCP-compatible client.

Tools

Tool Description
wais_discover Fetch a site's agents.json and show available actions
wais_register Register at a site using SD-JWT selective disclosure
wais_execute Execute any action (search, subscribe, purchase, etc.)
wais_confirm Confirm high-risk actions or complete payments (402 flow)
wais_status Check account status, plan, and credits

Quick Start

pip install wais-mcp
wais-mcp init

The setup wizard will ask you to pick your provider (Claude, ChatGPT, Cursor, etc.), choose project or global scope, and enter your API key. It writes the correct config file automatically.

Manual Setup by Provider

If you prefer to configure manually, all providers use the same wais-mcp command — only the config file location and format differ.

Claude Desktop

File: claude_desktop_config.json

{
  "mcpServers": {
    "wais": {
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Claude Code

File: .mcp.json (project root) or ~/.claude/settings.json (global)

{
  "mcpServers": {
    "wais": {
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

ChatGPT Desktop

Requires Developer Mode: Settings > Advanced Settings > Developer Mode.

File locations:

  • macOS: ~/Library/Application Support/ChatGPT/mcp-server-config.json
  • Windows: %APPDATA%\OpenAI\ChatGPT\mcp-server-config.json
  • Linux: ~/.config/ChatGPT/mcp-server-config.json
{
  "mcpServers": {
    "wais": {
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Restart ChatGPT after saving.

Gemini CLI

File: ~/.gemini/settings.json (global) or .gemini/settings.json (project)

{
  "mcpServers": {
    "wais": {
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Cursor

File: .cursor/mcp.json (project) or ~/.cursor/mcp.json (global)

{
  "mcpServers": {
    "wais": {
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Windsurf

File: ~/.codeium/windsurf/mcp_config.json

Or open from Windsurf: click MCPs icon in Cascade panel > Configure.

{
  "mcpServers": {
    "wais": {
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

VS Code (GitHub Copilot)

File: .vscode/mcp.json (project) or via Command Palette: MCP: Open User Configuration (global)

{
  "servers": {
    "wais": {
      "type": "stdio",
      "command": "wais-mcp",
      "env": {
        "PLATFORM_URL": "https://pod.deeger.io",
        "WAIS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Note: VS Code uses servers instead of mcpServers, and requires the type field.

Usage without MCP (Python SDK)

For custom agents, scripts, or any Python code — no MCP client needed:

import asyncio
from wais_mcp import WAISClient

async def main():
    client = WAISClient(api_key="your-api-key")

    # 1. Discover what the site offers
    site = await client.discover("https://serphub.deeger.io")
    print(site.name, site.list_action_ids())

    # 2. Register (shares only required claims via SD-JWT)
    await client.register(site)

    # 3. Execute actions
    result = await client.execute(site, "search", {"query": "python"})
    print(result)

    # 4. Confirm high-risk actions (if 402 returned)
    # result = await client.confirm(site, challenge_id)

    # 5. Check account status
    status = await client.status(site)
    print(status)

asyncio.run(main())

Works with OpenAI SDK, LangChain, CrewAI, or any Python agent framework.

How It Works

  1. Discover — Fetches /.well-known/agents.json from a site
  2. Register — Shares only required claims via SD-JWT selective disclosure
  3. Execute — Handles tokens, DPoP, endpoint resolution, and async polling automatically
  4. Confirm — Polls for completion after 402 confirmation challenges
  5. Status — Check credits and plan info

All authentication (PoD tokens, DPoP proofs) is handled transparently.

Configuration

Env Var Default Description
PLATFORM_URL https://pod.deeger.io WAIS Provider URL
WAIS_API_KEY (keychain fallback on macOS) API key from provider dashboard

Development

pip install -e ".[dev]"

License

MIT

from github.com/deegerhq/wais-mcp

Установка Wais

У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.

▸ github.com/deegerhq/wais-mcp

FAQ

Wais MCP бесплатный?

Да, Wais MCP бесплатный — установка в пару кликов через Unyly без оплаты.

Нужен ли API-ключ для Wais?

Нет, Wais работает без API-ключей и переменных окружения.

Wais — hosted или self-hosted?

Доступен hosted-вариант: Unyly запускает сервер в облаке, локальная установка не обязательна.

Как установить Wais в Claude Desktop, Claude Code или Cursor?

Открой Wais на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.

Похожие MCP

Compare Wais with

Не уверен что выбрать?

Найди свой стек за 60 секунд

Автор?

Embed-бейдж для README

Похожее

Все в категории ai