Aimarket Plugins

What alexar76/aimarket-plugins MCP server does

The alexar76/aimarket-plugins MCP server converts an AIMarket capability into packaging artifacts for local deployment. Its output is intended to make a capability usable as a self-hosted MCP product rather than only as a capability listed in the AIMarket ecosystem.

The package focuses on three generated outputs: a Docker image definition, an MCP manifest, and a Claude Desktop configuration file. These artifacts can be used to prepare a capability for a local MCP runtime or to configure an MCP client to start it.

The repository describes this component as one of 15 AIMarket Hub plugins. The other plugins cover areas such as safety, escrow, reputation, datasets, and streaming, but those components are not the tools exposed by this MCP server.

How it works

The server uses the official MCP Python SDK and communicates through stdio. An MCP client starts the Python server process, sends tool requests over the stdio transport, and receives the generated packaging results.

The main workflow is to submit an AIMarket capability to the packaging tool, then request the artifacts needed for deployment or client configuration. Docker-oriented output supports building a self-hosted image, while the manifest describes the MCP package and the Claude Desktop output provides an mcpServers configuration entry.

The alexar76/aimarket-plugins MCP server is therefore most relevant when an AIMarket capability needs to be exposed through a standard MCP client without manually creating these files. It is not described as a general-purpose gateway for the separate oracle server or for every plugin in the hub.

Setup and configuration

The README gives a direct source-run command from the repository root:

python plugins/aimarket-mcp-packager/mcp_stdio_server.py

The process uses stdio, so it should be launched by an MCP client or another MCP-compatible host that supports local stdio servers. The documentation specifically names Claude Desktop and Cursor, and states that other stdio MCP clients can also run it.

The README also states that Docker builds use the repository root and that all 15 plugins ship in the production hub image. The aimarket-mcp-packager PyPI project is marked as pending because of a PyPI new-project rate limit, so the documented source command is the supported setup path in the supplied material.

Tools and capabilities

The server exposes three tools:

• package_capability packages an AIMarket capability into the described MCP product artifacts.
• generate_dockerfile generates the Dockerfile used to build a self-hosted image.
• generate_claude_desktop_config generates a Claude Desktop mcpServers configuration.

Together, these tools cover packaging, container setup, and client configuration. The README also identifies an MCP manifest as part of the packaged output, although it does not list a separate manifest-generation tool by name.

Limitations and notes

The alexar76/aimarket-plugins MCP server is published in a read-only mirror. Pull requests made to this repository are overwritten during synchronization, and the repository identifies the AI-Factory monorepo as the canonical source.

The supplied material does not document environment variables, authentication settings, request schemas, or the exact input format for each tool. It also does not establish that the server itself invokes paid services or requires an API key. The separate aimarket-oracle-gateway is another MCP server and should not be treated as a tool included in this server.

The project is licensed under Apache-2.0 according to the repository badge.