MCP Inventory

Track, analyze, and optimize every MCP (Model Context Protocol) server and tool across your organization.

What is MCP Tracking?

MCP servers extend the capabilities of AI coding assistants by providing tools like file system access, database queries, API calls, and more. Intentra automatically detects and tracks MCP tool usage whenever scans capture AI coding sessions that invoke MCP servers.

The MCP Inventory tab (under Analytics) gives you full visibility into which MCP servers and tools your team is using, how often, and at what cost.

Key Features

Server Inventory

See every MCP server detected across your organization in a single view. Each server shows:

• Status — Active, Idle, or Dormant based on recent usage
• Tool count — Number of individual tools provided by the server
• Call count — Total invocations over the selected time period
• Last synced — When the server was last seen in a scan
• Estimated cost — Approximate cost attribution for usage
• ROI rating — High, Medium, or Low return on investment

Per-User Breakdown

Enterprise plans include a Users view that shows MCP usage broken down by team member, including:

• Servers and tools each user interacts with
• Individual call counts and cost attribution
• Top servers per user

Server Detail Pages

Click any server to see a detailed breakdown of its individual tools, including per-tool call counts, error rates, and cost.

Cost Attribution

Understand where your MCP spending goes. Intentra estimates costs per server and per tool so you can identify high-cost, low-value integrations and optimize your tooling spend.

Trust Score

Each MCP server receives a trust score when possible, based on source verification, community health, maintenance maturity, and usage reliability in your environment.

• Full Score (0-100) — Shown for servers with a verified public GitHub repository. Combines source code auditability, community adoption, project maintenance, and your org's error rate data.
• Partial Score (0-30) — Shown with a "Limited Data" badge for custom or private servers where no public source is available. Based solely on usage reliability in your environment.
• Source Unavailable — Shown when a previously verified repository has been archived or deleted.

Trust scores refresh automatically every 7 days. Scores appear on individual server detail pages and as badges on the MCP Inventory list.

How It Works

MCP tool usage is captured automatically during scans. When the Intentra CLI or extension detects MCP server calls in your AI coding sessions, it records the server name, tools invoked, and usage metadata.

Data flows through the standard Intentra pipeline:

1. Capture — AI coding session events include MCP tool calls
2. Normalize — Events are unified into Intentra's schema
3. Aggregate — MCP usage is rolled up per server, tool, and user
4. Display — The MCP Inventory tab visualizes the aggregated data

Availability

Getting Started

MCP tracking requires no additional configuration. Simply use MCP tools in your AI coding sessions with the Intentra CLI or extension installed. Data will appear on the Analytics > MCP Inventory tab once scans are synced.

For general setup, see Getting Started.