SuperAPI AI Directory

mcp-server

AgentBase: an open registry of agent instruction files. Search and retrieve system prompts, skills, workflows, domain packs and safety filters.

Open website

About this tool

Purpose

AgentBase MCP Server is an MCP server implementation of the AgentBase agent instruction registry. Its core purpose is to enable any AI agent compatible with MCP (Model Context Protocol) to directly access the AgentBase registry, search it, and retrieve instruction files from it. AgentBase is described as an open registry of agent instruction files, whose contents include system prompts, skills, workflows, domain packs, safety filters, and orchestration patterns.

Main Capabilities

This MCP server wraps the HTTP endpoints of the AgentBase registry and exposes them as structured MCP tools. In this way, AI agents can perform search and retrieval operations on the registry in the form of standardized MCP tool calls, without having to handle the underlying HTTP request details themselves.

Usage

The materials do not provide specific installation commands, configuration steps, or invocation examples. It is known that the server provides search and retrieval functionality through MCP tools, and any MCP-compatible AI agent can connect and use it. The complete registry contents can be browsed at agentbase-registry.vercel.app.

Applicable Scenarios

According to the materials, this server is suitable for the following scenarios:

  • AI agents that need to load domain-specific instructions at runtime;
  • Claude Code workflows that pull skills from a shared registry;
  • Teams that share system prompts and workflows across agents;
  • Safety filter retrieval in production-grade AI deployments.

Limitations and License

The project uses the CC0 license, and all instruction files can be used freely and without restriction. The materials do not provide specific information about fees, pricing, platform support scope, or performance results. This project is included in punkpeye/awesome-mcp-servers (marked 83k+ stars). The author is Chris Izworski, whose roles include solutions consultant at Prepared, former executive director of Michigan 911, and AI implementer, located in Bay City, Michigan. Related link: website chrisizworski.com,GitHub github.com/izworskic,Wikidata Q138283432。