Resources

Registry

The Registry is Procurator's unified catalog of all tools and resources available to agents in your organization. It aggregates the tool catalogs from all registered MCP servers into a single, searchable, governable inventory — making it easy to discover what's available, understand what's in use, and enforce policies on which tools agents may invoke.

Overview

As your Procurator deployment grows, tracking which tools are available, which agents use them, and what policies apply becomes a governance challenge. The Registry solves this by providing a single pane of glass over your entire tool ecosystem.

The Registry is auto-populated: when you register an MCP server, its tools automatically appear in the Registry. No manual entry required. The Registry augments the raw tool catalog with usage statistics, agent assignments, and governance metadata.

Registry vs MCP Servers

MCP Servers is the admin panel for managing connections to tool providers. The Registry is the catalog view of the tools those servers expose — with governance, usage data, and discoverability layered on top.

Control Panel

The screenshot below shows the live Procurator administration interface for this feature.

app.operativus.ai/procurator/registry
Procurator registry administration interface

Tool Registry — central catalog of all MCP tools with usage policies and agent assignments.

Key Capabilities

🔍

Unified Search

Search across all tools from all MCP servers by name, description, capability, or source server in one query.

📊

Usage Analytics

See which tools are called most, by which agents, with what success rate — identify high-value and underused tools.

🏷️

Tagging & Categorization

Add custom tags and categories to tools for discoverability — group tools by domain, risk level, or team ownership.

🚦

Tool Allow / Block

Block specific tools organization-wide or restrict them to specific teams without modifying the MCP server configuration.

🔗

Agent Assignments

View which agents have access to each tool and navigate directly to those agent configurations.

📋

Schema Documentation

Full JSON Schema documentation for every tool's input parameters and return types — always in sync with the live server.

Administration

Browsing the Registry

Navigate to Resources → Registry. The page presents a searchable, filterable table of all registered tools:

  • Search by tool name, description keywords, or source server name.
  • Filter by category using the tag sidebar (e.g., show only "Database", "Web", or "Code Execution" tools).
  • Sort by usage count, last called timestamp, or alphabetically.
  • Status filter: show Active, Blocked, or Deprecated tools.

Tool Detail View

Click any tool to open its detail view, which includes:

  • Description: What the tool does, provided by the MCP server.
  • Input Schema: Full JSON Schema of accepted parameters with types, descriptions, and required/optional designation.
  • Return Schema: The structure of the tool's return value.
  • Source Server: Which MCP server exposes this tool, with a link to the server's configuration.
  • Assigned Agents: A list of agents that have this tool's parent MCP server mounted.
  • Usage Statistics: Call count, success rate, p50/p95 latency over the last 7/30 days.
  • Recent Calls: A log of the most recent 20 invocations with input summaries.

Tool Governance

From the tool detail view, administrators can apply governance policies:

  • Block a tool: Prevents the tool from being called by any agent, even if the parent MCP server is assigned. Useful for deprecating dangerous tools without un-assigning the server.
  • Restrict to teams: Only allow specific teams' agents to invoke this tool. Applied as a runtime check during agent execution.
  • Require approval: Flag a tool as requiring an Approval Gate. Any workflow that invokes this tool without a prior approval gate is flagged with a warning.
  • Add tags: Add custom labels for categorization and policy targeting.
Blocking is not immediate for running agents

When you block a tool, agents that are currently executing and have already received the tool schema in their context may still call the blocked tool. The block takes effect on the next invocation. For immediate effect, terminate the running agent from the Admin Dashboard.

Permissions

  • registry:read— Browse the registry and view tool documentation
  • registry:govern— Block/unblock tools, restrict tool access, add governance tags