Official MCP Registry
Canonical open registry for publicly available Model Context Protocol servers.
Visionary Junction directory
Browse the Visionary Junction directory of MCP servers, APIs, agents, tools, data and specialist services. Compare pricing and access.
Canonical open registry for publicly available Model Context Protocol servers.
Discover, configure and connect hosted and local MCP servers.
Large MCP directory with server pages, code inspection and discovery tools.
Broad independent discovery catalogue for MCP servers and integrations.
Curated container-based MCP servers designed for consistent deployment.
Connect agents to repositories, issues, pull requests and GitHub workflows.
Open payment protocol for machine-readable, internet-native service payments.
Discovery surface for endpoints that can be purchased through x402.
Multimodal models and agent tooling for reasoning, creation and automation.
Claude models for reasoning, coding, research and long-context workflows.
Independent web search index available to agent and application workflows.
Models, datasets, inference endpoints and compute for open machine learning.
Serverless inference, fine-tuning and dedicated endpoints for open models.
Query, visualize and access blockchain data across many networks.
Enterprise blockchain data APIs and normalized datasets for applications.
Managed authentication and tool integrations for AI agents and applications.
A reusable instruction pack for AI coding agents working on React and Next.js performance, covering data fetching, bundle size, rendering and related implementation choices.
A local AI assistant with desktop and command-line interfaces that connects models to external tools through Model Context Protocol extensions.
An orchestration framework for stateful agent workflows, combining graph-based control flow with persistence, streaming and human-review steps.
Self-hosted gateway that gives applications an OpenAI-format interface to multiple language-model providers, with virtual keys and spend tracking.
Open-source chat application that brings multiple AI providers into one interface, with conversation branching and reusable presets.
A web-reading tool for MCP clients that turns retrieved HTML into Markdown and supports reading long pages in chunks.
Inspect MCP servers through a browser interface, a scriptable CLI or an interactive terminal client.
A Python SDK for typed agent loops, validated structured responses and function-based tools.
Build MCP servers and clients in Python, exposing functions as tools with generated schemas and validation.
A Python agent library with code-based actions, interchangeable models and MCP tool integrations.
Build agent applications and graph-based multi-agent workflows in Python or .NET, with checkpointing and human review patterns.
A hosted MCP connector that brings Exa web search, page fetching and authenticated research runs into compatible AI clients.
A coding agent for editor and terminal workflows, with reviewable file changes, command execution and configurable model providers.
Build tool-using agents, coordinate teams and workflows, and serve them through AgentOS APIs or MCP.
Terminal-based AI pair programming with repository mapping, Git integration and support for cloud or local language models.
Open-source AI coding extensions for VS Code and JetBrains, combining agent collaboration, code chat, in-file edits and inline completion.
Local-first testing and red teaming for LLM applications, with assertions, provider comparisons and browser-based results.
An MIT-licensed runtime for building with open models locally on macOS, Windows and Linux.
An MCP prompt that embeds a repository's staged diff and asks a model to draft a commit message.
Open-source browser agent that turns tasks into Python actions in a real browser, with local-browser and hosted-browser options.
Self-hostable memory engine for AI applications, available as a library or a server with a dashboard and per-user API keys.
Use MCP tools and prompts inside Zed's Agent Panel, with local or remote servers and configurable tool approval.
A unified model API for applications that need provider routing, fallback handling and a choice of language models without separate integrations for each provider.
Connect an MCP-capable assistant to Tavily's hosted web search, content extraction, crawling and research tools.
A desktop AI workspace combining chat, model management, MCP connectors and a local API server.
Clear the filters to see all directory listings.