MCP servers that give coding agents real developer tooling: linters, package registries, test runners, language docs and local build commands. 13,953 servers in this category.
Enables Claude Desktop to connect to Devin's SSE transport by reliably passing a Personal API Key for access to private repositories.
Tests MCP server authorization implementations against the 2026-07-28 specification using naive and hardened reference models.
Provides a reproducible laboratory for measuring communication overhead and latency across process and protocol boundaries in multi-agent architectures.
Converts browser HAR captures into MCP tools, allowing AI agents to replay authenticated web requests without needing an API key or SDK.
A minimal MCP server implementation in pure Python that provides JSON-RPC 2.0, SQLite memory tools, and hash-chained audit logging.
Provides an MCP server implementation featuring static, templated, dynamic, and binary resources with SSE subscriptions and cursor pagination.
Enables LLMs to load tabular files, profile data distributions, and execute sandboxed pandas queries for data analysis.
Provides a testing harness and evaluation suite for autonomous agents to detect production failures in tool-calling, loop control, and context management.
An open-source router that discovers, catalogs, and selectively exposes authorized MCP tools and skills to AI agents using Jev-based decision routing.
Provides a Python client, CLI, and REST gateway to interact with Robinhood's Agentic Trading MCP server for executing equity, option, or crypto orders.
Demonstrates how to build MCP tool servers and connect them to LangChain/LangGraph agents using stdio and streamable-http transports.
Provides a reference implementation for a secure enterprise healthcare MCP platform using synthetic FHIR data and multi-tier identity enforcement.
Provides a reproducible benchmark for evaluating agent-operated mobile device tooling using various MCP tool stacks and frozen tasks.
Provides a shared Linux workbench for coding agents to run builds, tests, and file processing with atomic resource reservations.
Measures the architectural cost of different MCP server design patterns on agent performance using success rates, turn counts, and token usage.
Provides AI agents with a source-backed memory layer using Git citations, local retrieval, and conversation memory for verifiable project knowledge.
Delegates Grok Bot coding tasks to local Claude Code and Codex CLIs via a custom MCP connector and Cloudflare tunnel.
Provides guides and a catalog of tools to securely connect AI agents to 1C:Enterprise and Bitrix24 using MCP, OData, and REST APIs.
An agentic Hyprland desktop operating system that lets AI agents operate the desktop through a permissioned MCP server.
Build multi-agent AI workflows and chatbots using pre-configured projects for document serving, Q/A bots, and orchestration.
