Model Context Protocol
Model Context Protocol (MCP) is an open-source standard for connecting AI applications to external systems. It standardizes how AI clients discover and use data sources, tools, prompts, and workflows so developers can build an integration once and make it available across compatible AI applications.
At a Glance
- AI application and agent developers
- Enterprise IT and platform teams
- Developer-tool and IDE companies
- Data, SaaS, and API providers building AI integrations
- +1 more
AI Tools by Model Context Protocol
(5)
MCP Registry
MCP Server Registry and Publisher
Model Context Protocol Servers
Reference MCP Server Examples
MCP Python SDK
Python SDK for MCP Servers
MCP Inspector
MCP Server Testing Tool
a2ui.me
MCP Apps Directory
Discussions
No discussions yet
Be the first to start a discussion about Model Context Protocol
Latest News
MCP 2026-07-28 specification released with a stateless protocol core, SDKs, and first-class extensions including MCP Apps.
Core maintainer update and continued work on MCP governance, roadmap priorities, and ecosystem stewardship.
Anthropic donated MCP to the Agentic AI Foundation under the Linux Foundation.
One-year anniversary release delivered the November 2025 MCP specification with tasks, authorization, enterprise security, extensions, and sampling improvements.
Products & Services
Open protocol defining standardized connections between AI applications (clients) and external data sources and tools (servers), including discovery, resources, prompts, tools, authorization, and transport behavior.
Official libraries for building MCP clients and servers, with maintained implementations including TypeScript, Python, Java, Kotlin, C#, Go, PHP, Ruby, Rust, and Swift.
Open-source reference and maintained server implementations that expose external systems and services to MCP clients; initial examples included Google Drive, Slack, GitHub, Git, Postgres, and Puppeteer.
Official open catalog and API for discovering and distributing publicly available MCP servers; it is designed as a primary source that public and private sub-registries can consume.
Market Position
MCP positions itself as a vendor-neutral interoperability layer for AI-to-tool and AI-to-data connections—often described as a USB-C-like standard for AI. Its main alternatives are proprietary or client-specific connector frameworks and competing agent/tool protocols; MCP differentiates through an open specification, multi-language SDKs, broad client adoption, community governance, and a central registry. Officially documented client support includes Claude, ChatGPT, Visual Studio Code, Cursor, MCPJam, Gemini, and Microsoft Copilot.
Leadership
Founders
David Soria Parra
Anthropic Member of Technical Staff and MCP lead maintainer; previously spent about a decade at Facebook/Meta working on developer tooling, source control, and static analysis. Earlier, he contributed to PHP, worked at Sun Microsystems optimizing PHP for SPARC, and served as PHP 5.4 and 5.5 release manager.
Justin Spahr-Summers
Anthropic engineer and software-development tools specialist; co-created MCP with David Soria Parra and helped drive its initial implementation. He is also known for open-source developer tooling work, including Sourcery and contributions to the Cocoa development ecosystem.
Executive Team
David Soria Parra
Lead Maintainer
MCP creator and Lead Maintainer; the project describes him as part of the team stewarding the protocol and its specification process.
Den Delimarsky
Lead Maintainer
Joined David Soria Parra as a Lead Maintainer in April 2026 to expand the project's leadership capacity.
Board of Directors
Founding Story
MCP began inside Anthropic in 2024 after engineers repeatedly encountered the same patterns for connecting AI applications to data and tools. David Soria Parra and Justin Spahr-Summers developed it to avoid a proliferation of bespoke, client-specific integrations; Anthropic open-sourced it on November 25, 2024 with reference servers and Claude Desktop support, inviting a broader community to extend it.
Business Model
Revenue Model
MCP is an open-source protocol and community project; official materials describe open specifications, SDKs, servers, and registry infrastructure rather than a paid subscription or usage-based revenue model.
Target Markets
- AI application and agent developers
- Enterprise IT and platform teams
- Developer-tool and IDE companies
- Data, SaaS, and API providers building AI integrations
- Open-source developers and protocol implementers
- AI assistants accessing calendars, knowledge bases, files, and business systems
- Enterprise chatbots querying multiple organizational databases
- Coding agents connecting to repositories, issue trackers, design tools, and development environments
- Automating payment, API testing, model-management, and dataset-search workflows
- AI applications invoking search engines, calculators, specialized prompts, and domain tools
- Interactive applications embedded in AI clients
- Block
- Apollo
- Zed
- Replit