MCP servers for building, orchestrating and running autonomous AI agents, multi-agent systems and LLM tooling. Each entry is scanned with the deterministic Capability-Flow Trust Model — grades are computed, never self-reported.
Tools for writing MCP clients and servers without pain
MCP server for SAPUI5/OpenUI5 development
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Real-time web search, reasoning, and research through Perplexity's API
MCP nodes for n8n
Model Context Protocol (MCP) server for freee API integration
Browser transport implementations for Model Context Protocol (MCP) - postMessage, Chrome extension messaging, and iframe communication for AI agents and LLMs
Model Context Protocol server for Forest Admin with OAuth authentication
An MCP interface into the Bright Data toolset
Authenticated MCP transport with HTTP Signatures for AAuth agents
I am Legion, for we are many ~ MCP-native LLM councils for debates, juries, blind panels, refinement, and custom multi-model deliberation.
TypeScript MCP 1.x server for Canvas LMS — 117 tools across Canvas courses, assignments, grades, gradebook history, quizzes, New Quizzes (LTI), outcomes, discussions, admin workflows, and more.
MCP Server for Paddle Billing
MCP server for Packkit — let AI agents scaffold modern npm packages, CLIs, services, and apps as a native tool.
MCP server for interacting with Autodesk Revit through AI assistants like Claude
Edit Word (.docx) and OpenDocument (.odt) files with tracked changes, redlines, and formatting preservation. Built for AI coding agents. Apache-2.0 licensed, 100% local processing.
A collection of tools for Twist using AI
Enable Cursor, VS Code, Claude Code or any MCP-enabled IDE to help you vibecode and manage Home Assistant: create automations, design dashboards, tweak themes, modify configs, and deploy changes using natural language
Czech & EU due diligence in one call — company facts, insolvency (ISIR), EU+OFAC sanctions, VAT reliability, risk scoring (0-100), and statutory chain. EU coverage via GLEIF/LEI. Official state registries only, no Cribis/Bisnode reselling.
Programmatic add/link/unlink for MCP servers across 23 AI coding agents (Claude Code, Claude Desktop, Cursor, VS Code, Codex, Gemini CLI, Zed, Cline, OpenCode, Goose, Kiro, Windsurf, and more). Functional API with dry-run support.
Model Context Protocol (MCP) server for OpenClaw AI assistant integration
🌐 YYC³ Production-Ready Internationalization (i18n) Framework - High-performance, plugin-based, zero-dependency i18n solution for TypeScript/JavaScript applications
MCP server exposing public instruction workflows as tools, backed by hidden AI agent skills for requirements, orchestration, quality, research, evaluation, governance, resilience, and physics-inspired analysis
MCP server enabling AI assistants to securely execute SSH commands, transfer files via SFTP, manage port forwarding, and use parameterized command templates with comprehensive security controls
AI & Agents MCP servers expose agent frameworks, orchestration engines and LLM tooling to AI assistants through the Model Context Protocol. Instead of a single chat loop, an assistant such as Claude or Cursor connected to one of these servers can create, coordinate and monitor autonomous agents: orchestrators in the LangChain, CrewAI or AutoGen ecosystems, vector memory stores, model gateways and evaluation harnesses.
Three uses dominate in practice. An assistant can spawn and coordinate sub-agents, splitting a research or coding task across specialized workers and merging the results. It can keep persistent agent memory, writing state to a vector store and recalling it across sessions. And it can route work through model gateways for batch inference, tool-call chains or automated evaluations. Because agent tooling usually wraps other services, these servers often bundle capabilities you would otherwise find under API Development or Files & Storage: HTTP clients, credential handling, workspace file access.
The security risks in this category are structural, because an orchestration server sits inside every agent loop. A malicious or over-permissioned one can read full conversation context, inject instructions into delegated agents, burn your LLM provider keys, or chain tool calls that end in code execution and silent outbound traffic. Prompt injection also amplifies here: one poisoned tool description can propagate to every sub-agent it touches. Before installing:
A hostile or careless one can: agent orchestration servers routinely see provider keys, conversation context and tool outputs, which is exactly what exfiltration needs. Stick to servers whose Trust Score, capability profile and findings you have actually reviewed, and grant the narrowest permissions your workflow allows.