Announcing vMCP 0.5.0 and opensource release
We're excited to introduce virtual MCPs and announce the open-source release of vMCP 0.5.0. Making AI workflows and agent development more accessible and powerful. Today, vMCP is fully open source under the MIT license, empowering non-developer, developers and organizations to build sophisticated AI workflows with MCPs without vendor any lock-in. Customise, Compose and extend MCPs for your AI needs.
Dependable AI tasks need goals, planning, reasoning, tool-use and context engineering. vMCPs are our first step to build useful AI assistants and agents by improving tool-use and context engineering
What is vMCP?
vMCP (virtual Model Context Protocol) is an AI configuration and management platform built on top of the Model Context Protocol. Lego blocks for AI workflows and agents, allowing you to compose, customize and extend multiple MCPs into powerful, reusable AI components. Connect vMCPs to your AI clients (Claude, Chatgpt, VSCode) or to your agents and get the power and configurability of vMCPs.
vMCP is both
- a specification that builds on and extends MCPs vMCP.json
- a platform to create and deploy vMCPs
vMCP Architecture
Think of vMCP as a drop-in replacement for MCPs, but with superpowers:
- Configure once, use everywhere across all your AI clients and agents
- Mix and fine-tune multiple MCPs for specific use cases with precision
- Share and reuse vMCPs across your team and the community
Why We Built This
The Model Context Protocol has unlocked incredible possibilities for AI integrations, but users and developers quickly hit limitations:
- Configuration Hell: Managing MCP configs across multiple clients (Claude, ChatGPT, VSCode, Cursor, Gemini) is tedious
- Auth: Each mcp client needs its own auth for all the MCPs. vMCPs gives a standard auth for all MCPs
- Lack of Customization: Can't modify or extend existing MCPs for specific workflow needs
- No Composition: Building complex workflows requires piecing together multiple tools manually in code
vMCP solves these problems by providing a layer of abstraction and a no-code configuration interface on top of MCPs.
Core Features in 0.5.0
1. Unified Auth and Configuration
Stop managing separate MCP configurations for every AI client. Configure your vMCP once, and it works seamlessly across Claude Desktop, ChatGPT, custom agents, and any MCP-compatible client. - one auth and everything is ready across clients
2. Powerful Prompt Syntax for Context Engineering
One of the most powerful features is to extend upstream MCPs with custom prompts tuned to your wworkflows. Our advanced prompt syntax gives you fine-grained control over your AI workflows:
- Tool Calling: Explicitly invoke tools with parameters, get tool output in the context
- Resource Embedding: Inject context from files, APIs, or databases
- Dynamic Configuration: Adapt behavior based on vMCP config variables or prompt parameters
@config.VAR_NAME - Config variables
@param.name - Context variables
@tool.server.name() - Tool calls
@prompt.server.name() - Prompt execution
@resource.server.name - Resource references
Example dynamic prompt:
The current time is @tool.AllFeature.get_current_time(timezone_name: str = "UTC")
User is at @tool.AllFeature.get_location(city: str = "@config.CITY")
Book a table with AllFeature.book_table() for @param.party_size 1 hour from now
3. Web Tools - Any API as a Tool
Transform any web API into an AI-callable tool in minutes. No coding required for simple integrations:
- REST APIs
- GraphQL endpoints
- Webhooks
- Custom authentication
4. Python Tools - extend MCPs with custom tools with code
Execute Python code in a secure container. Ideal for complex data processing, calculations, and logic.
5. Fine-Tuned Tool Control
Customize existing MCP tools for specific use cases:
- Filter which tools are exposed
- Modify tool names and descriptions for better AI understanding
- Set default parameters
- Add validation rules
6. Community vMCPs (1xn hosted version)
Package your AI configurations as reusable vMCPs:
- Discover and use communtity vMCPs
- Share with your team or the community
- Build on top of others vMCPs
Use Cases
| Personal | Business and Enterprise |
|---|---|
| Research assistant with access to your notes and bookmarks | Rapid Iteration: Build and deploy agents in hours, not weeks |
| Code reviewer integrated with your GitHub repos | Customer Personalization: Create unique AI experiences for each customer |
| Personal productivity agent connected to your calendar and todo list | Security & Audit: Centralized control over AI tool access and usage |
| Make custom vMCPs for your work automations | Team Collaboration: Share configurations across development teams |
1xn-vmcp : The Hosted vMCP Platform
While vMCP is fully open source, we also offer 1xn, a hosted platform with enterprise features:
- Multi-user Support: Team collaboration and role-based access control
- Managed Infrastructure: We handle hosting, scaling, and monitoring
- Enterprise Security: SOC2 compliance, SSO, audit logs (coming soon...)
- Premium Support: Dedicated support and onboarding assistance
Whether you self-host vMCP or use 1xn, you're working with the same powerful core technology.
What's Next: Our Roadmap
We're just getting started. Here's what's coming:
Completions API/SDK
A fully-featured completions API with native vMCP integration:
- Use vMCPs natively in your AI SDKs
- Built-in support for multi-step workflows
Agents with Triggers
Deploy autonomous agents that run on schedules:
- Cron-based execution
- Event-driven triggers
- Background processing
Agent Shells
Connect your agents to communication platforms (starting with Slack):
- Slack integration (coming soon)
- WhatsApp business integration
- Email support
- Discord and Teams support
Our vision: vMCPs → Agents → Shells - a complete stack for building and deploying AI solutions.
Getting Started
uvx --from 1xn-vmcp vmcp run
Check out our documentation for more options
Use 1xn Hosted Platform
Visit 1xn.ai to sign up for a free account (no card deatils required) and start building and sharing your vMCP creations immediately.
Open Source
vMCP is MIT licensed - use it however you want, commercially or personally.
We welcome contributions:
- Report bugs and request features on GitHub
- Submit pull requests
- Share your vMCP configurations with the community
- Help improve documentation
Join the Community
- GitHub: github.com/1xn-labs/1xn-vmcp
- Documentation: Read the docs
- Discord: Join our community (coming soon)
- Twitter: Follow @1xn_ai for updates
Acknowledgments
We want to give a huge shoutout to Anthropic and the entire Model Context Protocol community for creating such an incredible foundation. MCP has fundamentally changed how we think about AI integrations
vMCP is built on top of MCP, extending its capabilities while maintaining full compatibility. We're excited to contribute back to this ecosystem and help push the boundaries of what's possible with context-aware AI.
A huge thank you to everyone who provided feedback during our closed beta. Your insights have been invaluable in shaping 1xn vMCP into what it is today.
We're excited to see what you'll build with vMCP. Let's make AI more accessible, composable, and powerful - together.
Ready to get started? Check out the documentation or try 1xn today.
Star us on GitHub if you find vMCP useful!