HVTracker MCP

What HVTracker MCP does

HVTracker MCP MCP server gives AI clients access to public trust information for AI agents, frameworks, MCP servers, packages, and GitHub repositories. Its main use is pre-connect checking: an agent can inspect a target before adding it to a workflow or dependency set.

The server can return a trust verdict for an identified MCP server, package, repository, or agent. It can also retrieve a tracked agent’s trust profile, including MCP status, providers, plugin surface, provenance drift, and the URL of its Ed25519-signed trust credential. For comparisons, it places two agents side by side and returns an evidence-based verdict with a published comparison-page link.

How it works

Clients call the MCP tools through either the hosted endpoint at https://hvtracker.net/mcp or a locally installed stdio process. Requests query HVTracker’s public registry rather than requiring a private service credential.

The available operations cover both individual targets and broader inventories:

• Search the registry by agent name, repository, description, or category.
• Submit a requirements file, package manifest, MCP client configuration, or plain list to scan a dependency stack.
• List registry categories and their agent counts.
• Retrieve top tracked agents and MCP servers by HVTrust score, optionally within a category.
• Review 90-day trust-score, grade, and rank history for one tracked agent.

Stack scans return a verdict for each submitted item and a summary for the overall set. The history tool is intended to show whether a tracked agent’s trust position is improving or declining over time.

Setup and configuration

For a remote MCP configuration, point the client at:

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

For local stdio use, install the package from npm with npm install -g hvtracker-mcp or from PyPI with python3 -m pip install hvtracker-mcp. The local client entry uses the hvtracker-mcp executable:

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

Development environments can override the registry endpoint with HVTRACKER_BASE_URL, such as a local test service. Normal operation does not require an API key.

Tools and capabilities

HVTracker MCP MCP server exposes these tools:

• verify_mcp_server checks a target before connection.
• check_agent_trust returns a tracked agent or framework profile.
• compare_agents compares two agents and provides an evidence-based result.
• search_agents searches registry records.
• scan_stack checks multiple dependencies and summarizes the stack.
• list_categories returns categories and counts.
• get_leaderboard returns ranked agents and MCP servers.
• get_agent_history returns 90-day score, grade, and rank history.

Limitations and notes

The service only describes data available in the public HVTracker registry. A target must be identifiable through the values accepted by the relevant tool, and an untracked agent may not have a trust profile or history available.

HVTracker MCP MCP server sends the user-provided search string or server identifier to https://hvtracker.net to obtain public trust data. It does not require an API key and does not write to user systems. Current registry data and scoring methodology are maintained by HVTracker, so developers should consult that service when interpreting results.

Getting started with this HVTracker MCP MCP server

Always refer to the official documentation for the most accurate and up-to-date information.