The Network Catalog (webGCP)

Semantic search over The Network catalog (skills, agencies, roles) with signed webGCP provenance.

我該用這個嗎

品質與安全性

A
說明品質
100%
結構描述完整度
100%
命名品質
90%
汙染風險
100%
權限相符程度
100%
協定合規性
100%

根據工具定義與協定合規性的自動化分析。

上下文成本

~232Token(工具定義)
~831 B典型回應大小
極小的注意力影響(128k 上下文的 0.18%)

這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高,可用於其他工作的注意力就越少。

安裝

一鍵安裝

將以下內容加入你的 `claude_desktop_config.json` 檔案:

{
  "mcpServers": {
    "catalog": {
      "url": "https://catalog.myparallel.dev/mcp"
    }
  }
}

遠端端點

https://catalog.myparallel.dev/mcpstreamable-http
https://catalog.myparallel.dev/ssesse

它能做什麼

工具清單

工具(2)

🟢 唯讀🟡 寫入🔴 刪除⚪ 未知
🟢local_search(query, entity_type, limit)

Semantic search over The Network catalog. Returns skill entities ranked by embedding similarity.

輸入結構描述

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "Natural-language search query."
    },
    "entity_type": {
      "type": "string",
      "description": "Entity type to search. Default: skill.",
      "enum": [
        "skill",
        "agency",
        "sector",
        "role"
      ]
    },
    "limit": {
      "type": "number",
      "description": "Max results (1–50). Default: 10."
    }
  },
  "required": [
    "query"
  ]
}
🟢read_surface(url, max_chars)

Read an external web surface, discovery-first: returns its WebMCP/MCP server-card manifest when one exists; otherwise DOM-dehydrates the page into an indexed interactive-element text map (static parse, truth_class=inferred).

輸入結構描述

{
  "type": "object",
  "properties": {
    "url": {
      "type": "string",
      "description": "Absolute http(s) URL of the surface to read."
    },
    "max_chars": {
      "type": "number",
      "description": "Text-map character budget (default 8000)."
    }
  },
  "required": [
    "url"
  ]
}

社群

為此伺服器評分

證據

近期觀測

已驗證未記錄版本2 個工具
已驗證未記錄版本2 個工具
已驗證未記錄版本2 個工具
已驗證未記錄版本2 個工具
已驗證未記錄版本2 個工具
已驗證未記錄版本2 個工具