Packages

Provides an MCP server that exposes Utsunomiya City open data via natural-language queries and MCP tools (list datasets, get dataset schemas, query dataset contents, and analyze datasets with caching).

Provides an MCP server integration for Zed that exposes a `convert_to_markdown(uri)` tool to convert documents (PDF/DOCX/XLSX/PPTX), images (with OCR), audio (with transcription), web pages (URLs), structured data (CSV/JSON/XML), and ZIP archives into AI-consumable Markdown context via MarkItDown.

OpenMetadata is a unified metadata platform for data discovery, observability, and governance. It provides a central repository for cataloging data assets across 84+ connectors, with built-in lineage tracking, data quality testing, and collaboration tools.

AgentCrew is a Python multi-agent AI framework and desktop/CLI application for building teams of specialized agents that collaborate via transfer or parallel delegation. It supports multiple model providers and tool integrations, including MCP-connected external tools, and can expose agents as HTTP services via an A2A (JSON-RPC) server.

AgentScope is a Python multi-agent framework for building, orchestrating, and running LLM agents with reusable abstractions (agents, toolkits, memory, planning/workflows) and integrations including MCP tool support and additional protocols (e.g., A2A) plus observability via OpenTelemetry.

DevDuck is a Python-based, “one-file” self-healing AI agent runtime that can hot-reload code, run many built-in tools (including shell/git/browser/messaging), and connect across multiple interfaces (CLI/TUI/WebSocket/TCP) and networks (Zenoh P2P + a unified mesh/relay). It can also expose itself as an MCP server and deploy to AWS AgentCore.

Provides a headless Ghidra server that exposes Ghidra reverse-engineering workflows to AI agents via the MCP (Model Context Protocol), with stdio (default) and TCP transports and both real (pyghidra) and fake-backend modes.

Provides a local, read-only MCP server that exposes Google Search data as relational tables via CData's JDBC Driver. The MCP server runs over stdio so an MCP-capable desktop client (e.g., Claude Desktop) can query live Google Search data using tool calls like get_tables, get_columns, and run_query (SQL SELECT).

Provides an MCP (Model Context Protocol) server for the Spiral Framework with automatic tool discovery via PHP attributes, schema generation/validation for tool inputs, configurable transport (HTTP/stream/STDIO), middleware support, and optional session handling.

A Python Model Context Protocol (MCP) server suite that exposes Huawei Cloud services as MCP tools, enabling AI agents to manage cloud resources (e.g., ECS, OBS, GaussDB and many other Huawei Cloud products) through standardized MCP transports such as http/sse/stdio.

A spec-driven Go template for building Model Context Protocol (MCP) servers. It provides default tools/resources/prompts and can be configured at startup via a JSON spec (v1) that controls tool names/inputs, resources, prompts, and a catalog dataset. It supports stdio (default) and an HTTP transport with documented REST-style endpoints.

An MCP server that exposes Oracle Cloud Infrastructure (OCI) Core Networking (VCN) service APIs, enabling AI agents to create and manage Virtual Cloud Networks, subnets, route tables, internet gateways, NAT gateways, security lists, network security groups, and other foundational OCI network resources.

parse-server is a self-hosted Parse Server implementation that provides a Parse-compatible backend (REST API and real-time features depending on setup) for Parse SDK clients. It works as a server process connected to a database (commonly MongoDB) and exposes endpoints for data models, queries, users, sessions (depending on configuration), file uploads, and related Parse behaviors.

A Model Context Protocol (MCP) tool server that connects to a PostgreSQL database (via DSN) and exposes MCP-callable tools to introspect schema (schemas/tables/columns/indexes/foreign keys) and run SQL for DQL/DML/DDL/DCL operations within transactions (as described).

Provides a local, read-only MCP server that exposes Quickbase data as relational tables using the CData JDBC Driver for Quickbase. It runs as a stdio-based MCP server (intended for local clients like Claude Desktop) and offers tools to list tables/columns and run SQL SELECT queries via JSON-RPC tool calls.

Provides a locally hosted, read-only MCP server that exposes SAP Business One data (via CData JDBC Driver for SAP Business One) to MCP-capable clients (e.g., Claude Desktop). It wraps the JDBC driver so the client can list tables/columns and run SQL SELECT queries through MCP tools.

Provides an MCP (Model Context Protocol) server that lets MCP clients interact with SingleStore Management API to manage organizations/workspaces, operate on starter workspaces (create/resume/terminate), run SQL on a workspace, and manage SingleStore Spaces notebook files and jobs.

Sonarr is a self-hosted media management application for organizing and downloading TV series. It manages show tracking, episode/file downloads via external download clients, and automates workflows based on configurable rules.

MCP (Model Context Protocol) server that exposes a simple “to-do list”/task-management capability to AI agents via MCP tools (create/list/update/delete tasks).

1mcp is a central proxy/hub that aggregates multiple upstream MCP servers and exposes a unified streamable-http MCP interface to downstream clients. It discovers upstream tools/resources/prompts, prefixes capability names (e.g., serverName___toolName), routes downstream requests to the correct upstream server, and supports dynamic hot-reloading of the upstream server configuration with live capability updates.