Skip to main content

Beak Hub Overview

Beak Hub is the catalog and marketplace layer of the Beak AI platform. It sits under the Hub · Marketplace eyebrow and gives every team member a single, curated destination for pre-built MCP servers, drop-in agents, and packaged agent suites. Rather than assembling integrations by hand, teams browse the catalog and connect what they need, with consistent configuration and credential storage behind it. Agents and suites are then installed into a Runtime from that Runtime's Marketplace.

The Hub dashboard opens with the tagline: "Pre-built MCP servers and agents, ready to connect. Connect MCP servers to any client and deploy agents into your Runtime — one catalog, consistent config, no custom integration." From there four sections give an at-a-glance picture of the catalog and quick paths to the three catalog areas.

Note: That tagline is legacy copy from when Hub pushed agents into a Runtime. It no longer describes how installation works — Hub is the catalogue, and the Runtime installs from it. See Marketplace.

KPI Row​

Four stat tiles at the top of the dashboard summarise the catalog at a glance:

  • MCP Servers — Total number of published MCP server entries in the catalog.
  • Agents — Total number of published agent entries in the catalog.
  • Tools · Capabilities — Aggregate count of tools across all MCP servers and capabilities across all agents, displayed as two numbers separated by a dot.
  • Categories — Count of distinct categories in use across both the MCP server and agent catalogs combined.

The Featured MCP servers panel lists up to four MCP server entries from the catalog. Each row shows the server name, tagline, and a badge with the total tool count for that server. Clicking any row navigates to the full MCP server detail page. A View all action at the top right of the panel jumps to the complete MCP Servers catalog.

The Featured agents panel mirrors the featured servers layout and lists up to four agent entries. Each row shows the agent name, tagline, and a badge with the total capability count. A View all action navigates to the full Agents catalog.

Category Breakdown​

The By category — MCP servers + agents distribution panel shows how catalog entries are spread across categories. Each row displays a category name, a split progress bar, and a summary of the form N mcp · N agent · N%. The solid portion of the bar represents MCP servers in that category; the lighter portion represents agents. Categories are ranked by total entry count, highest first.

Two primary actions appear in the hero strip:

  • MCP servers — Navigates to the full MCP Servers catalog.
  • Agents — Navigates to the full Agents catalog.

The Hub sidebar also provides direct links to MCP Servers, Agents, Agent Suites, and — for platform administrators — Tenants & API Keys.

Best Practices​

  • Use the KPI row as a quick health check — if the categories count is low, the catalog may not yet be fully seeded across all use-case domains.
  • The Featured panels surface the first entries alphabetically; use the full catalogs with the search and category filter to find specific servers or agents.
  • Install agents and suites from the Marketplace in the target Runtime. Hub is the catalogue; the Runtime pulls from it.
  • Category badges on cards are consistent across MCP Servers and Agents, so a productivity server and a productivity agent will share the same accent color, making cross-catalog discovery easier.
  • MCP Servers — Browse and connect pre-built MCP server entries.
  • Agents — Browse the pre-built agent catalogue; install from the Runtime's Marketplace.
  • Agent Suites — Install curated bundles of agents, skills, datasources, and MCP servers.
  • Tenants & API Keys — Register tenants and manage their Hub API keys.