HVTracker MCP

io.github.YugantM由 GitHub 帳號 YugantM 發布v0.4.0更新於 Nov 10, 2026

Pre-connect trust checks for AI agents and MCP servers using HVTracker's public trust registry.

已驗證Streamable HTTP可網頁執行開發者工具安全與監控AI 與機器學習

概覽

AI 產生的概覽

讓助理查詢 HVTracker 的公開信任登錄檔,在連線 AI 代理、MCP 伺服器和相依套件之前進行信任檢查。

可以做什麼
提供工具,為 MCP 伺服器、套件、GitHub 儲存庫和代理名稱回傳連線前的信任判定,並為已收錄的 AI 代理和框架提供信任設定檔,包括執行階段能力和已簽章信任憑證的 URL。它還能並排比較兩個代理、搜尋登錄檔、依 requirements.txt、package.json 或 MCP 用戶端設定批次檢查整組相依套件、列出分類、依 HVTrust 分數顯示排行榜,以及回傳某個代理 90 天的信任分數歷史。
適用情境
適合在加入 AI 代理、MCP 伺服器或套件相依項目之前需要信任訊號的情況,也適合在 CI 中審查相依性變更。用來比較代理,或追蹤某個已收錄代理的信任分數是上升還是下降,也很方便。
執行需求
可作為託管遠端端點使用,網址為 npm 套件 hvtracker-mcp 或 PyPI 套件 hvtracker-mcp 在本機以 stdio 方式安裝。未宣告需要 API 金鑰或環境變數;需要能連線到 hvtracker.net。README 提到可選用 HVTRACKER_BASE_URL 變數來測試其他基礎網址。
安裝前請注意
搜尋字串以及伺服器或代理識別碼會傳送到 以取得公開信任資料,因此查詢內容會離開本機。README 聲明不需要 API 金鑰,也不會寫入使用者系統。信任判定來自第三方登錄檔,依賴之前請到 HVTracker 網站核實方法與目前資料。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 HVTracker MCP,將其新增到工作區。
  2. 為需要使用其工具的對話啟用該服務。

Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.YugantM%2Fhvtracker-mcp/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "hvtracker": {
      "type": "http",
      "url": "https://hvtracker.net/mcp"
    }
  }
}

README

HVTracker MCP

[Smithery] [npm] [PyPI] [License: MIT]

MCP server for checking supply-chain trust before connecting to AI agents, frameworks, or MCP servers.

The hosted remote server is:

json
{  "mcpServers": {    "hvtracker": {      "url": "https://hvtracker.net/mcp"    }  }}

This repository also provides a local stdio package for clients that prefer package-based installation.

Tools

  • verify_mcp_server: pre-connect trust verdict for an MCP server, package, GitHub repo, or agent name.
  • check_agent_trust: trust profile for a tracked AI agent or framework — incl. runtime capabilities (MCP status, providers, plugin surface, provenance drift) and the URL of its Ed25519-signed trust credential.
  • compare_agents: two agents side by side with an evidence-based verdict and the published compare-page link.
  • search_agents: search the HVTracker registry by name, repo, description, or category.
  • scan_stack: bulk pre-connect trust check for a whole dependency set — paste a requirements.txt, package.json, MCP client config, or a plain list and get a trust verdict per item plus a stack summary.
  • list_categories: list the HVTracker categories with agent counts, so you can then pull a category's leaderboard.
  • get_leaderboard: top tracked AI agents and MCP servers ranked by HVTrust score, optionally filtered to one category.
  • get_agent_history: 90-day trust-score, grade, and rank history for one tracked agent — is it improving or declining?

Check your AI dependencies in CI

HVTrust Gate runs the same trust checks in GitHub Actions. With no configuration it reads your requirements.txt, pyproject.toml, package.json and MCP client configs, and comments on pull requests that change them with the HVTrust grade of each AI agent and MCP server you depend on.

yaml
name: HVTruston: [pull_request]permissions:  contents: read  pull-requests: writejobs:  trust:    runs-on: ubuntu-latest    steps:      - uses: actions/checkout@v7      - uses: YugantM/hvtrust-gate@v1

Local Install

With npm:

bash
npm install -g hvtracker-mcp

With PyPI:

bash
python3 -m pip install hvtracker-mcp

Example MCP client config:

json
{  "mcpServers": {    "hvtracker": {      "command": "hvtracker-mcp"    }  }}

Development

bash
python3 -m pip install -e ".[dev]"python3 -m pytesthvtracker-mcp

Use a different HVTracker base URL while testing:

bash
HVTRACKER_BASE_URL=http://localhost:8080 hvtracker-mcp

Registry Publishing

The official MCP Registry manifest is server.json.

bash
mcp-publisher login githubmcp-publisher publish

In GitHub Actions, run the "Publish MCP Registry" workflow after the npm, PyPI, and GHCR packages for the same version are live.

The server name is:

text
io.github.YugantM/hvtracker-mcp

Claude Desktop Extension

Tagged releases build an .mcpb bundle for Claude Desktop from manifest.json. To build it locally:

bash
npm ci --omit=devnpx @anthropic-ai/[email protected] pack

Privacy

HVTracker MCP sends the user-supplied search string or server identifier to https://hvtracker.net to fetch public trust data. It does not require an API key and does not write to user systems. See the HVTracker site for current data and methodology, and see PRIVACY.md for the repository privacy note.

來源:README.md,提交 d1dd8e4

工具

8

check_agent_trust

Check Agent Trust

Get the HVTracker supply-chain trust profile for a tracked AI agent or framework.

輸入參數name_or_repo必填

verify_mcp_server

Verify MCP Server

Pre-connect trust verdict for an MCP server, package, GitHub repo, or agent name before connecting an AI agent to it.

輸入參數server必填

compare_agents

Compare Agents

Compare two tracked AI agents side by side: trust scores, grades, runtime capabilities, and an evidence-based verdict.

輸入參數a必填b必填

search_agents

Search Agents

Search tracked AI agents and frameworks by name, repo, description, or category, ranked by trust score.

輸入參數limitquerycategory

scan_stack

Scan Stack

Bulk pre-connect trust check for a whole dependency set — paste a requirements.txt, package.json, MCP client config, or a plain list and get a trust verdict per item plus a stack summary.

輸入參數input必填

list_categories

List Categories

List the HVTracker categories with agent counts, so you can then pull a category's leaderboard.

無需輸入參數。

get_leaderboard

Get Leaderboard

Top tracked AI agents and MCP servers ranked by HVTrust score, optionally filtered to one category.

輸入參數limitcategory

get_agent_history

Get Agent History

90-day trust-score, grade, and rank history for one tracked agent — is it improving or declining?

輸入參數name_or_repo必填

版本歷史

2
  1. v0.4.0最新Sep 29, 2026
  2. v0.3.0Sep 16, 2026

評分

已滿足適用檢查項的 100%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具必需15 分
  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分