AgentxSuite – Open-Source Control Plane for AI Agents Using MCP

I’ve been working on AgentxSuite, an open source control plane for AI agents built around the Model Context Protocol (MCP).

Most teams building agent features run into the same issues:

each agent needs its own tool wiring

permissions, policies and audit logs scatter across different services

no central place to register tools, resources or prompts

hard to track what an agent actually did across runs

integrating with external MCP servers (Postgres, filesystem, etc.) is repetitive

AgentxSuite tries to solve this by giving you a single layer to manage:

agents

tools

resources

prompts

policies

audit trails

token and usage tracking

multiple MCP servers (local or remote)

It also includes an Agent Designer Canvas to visually inspect the full graph of agents, their tools and their policies.

If you want to integrate MCP into your own product, or if you’re exploring multi-agent architectures with proper access control, feedback welcome. https://github.com/alparn/agentxsuite