1xn-vmcp
by 1xn-labs
Overview
An open-source platform for composing, customizing, and extending multiple Model Context Protocol (MCP) servers into a single logical, virtual MCP server, enabling fine-grained context engineering for AI workflows and agents.
Installation
uvx --from 1xn-vmcp@latest vmcp runEnvironment Variables
- VMCP_DATABASE_URL
- VMCP_HOST
- VMCP_PORT
- VMCP_LOG_LEVEL
- VMCP_DEBUG
- VMCP_BASE_URL
- VMCP_DUMMY_USER_ID
- VMCP_DUMMY_USER_EMAIL
- VMCP_DUMMY_USER_TOKEN
- VMCP_ENABLE_TRACING
- VMCP_OTLP_ENDPOINT
- VMCP_TTL_SECONDS
- VMCP_CLEANUP_EVERY_SECONDS
- VMCP_FRONTEND_PATH
- VMCP_PROJECT_ROOT
- VMCP_DOCS_PATH
- GITHUB_CLIENT_ID
- GITHUB_CLIENT_SECRET
Security Notes
This server has critical security vulnerabilities that make it unsafe to run in any publicly exposed environment, and potentially unsafe even locally if untrusted users can define custom servers or tools. 1. **Arbitrary Code Execution via stdio MCP servers:** The `MCPClientManager` allows users to configure and execute arbitrary commands on the host system via `stdio` transport type MCP servers. While input validation exists for format and length, it does not prevent the execution of malicious commands. 2. **Arbitrary Python Code Execution via Custom Python Tools:** The `python_tool` feature explicitly executes user-provided Python code using `exec`. Although it attempts sandboxing with `RestrictedPython`, the codebase itself acknowledges this is 'not a complete security solution against determined attackers.' This is a severe, known vulnerability. 3. **Hardcoded Dummy Credentials and Default Open Access:** In OSS mode, it uses hardcoded dummy user credentials (`local-user`, `local-token`) and enables overly permissive CORS (`allow_origins=['*']`). This means any instance running in OSS mode is unauthenticated and fully accessible to anyone who knows these public default values, making it highly vulnerable if exposed to the network. 4. **Sensitive Data Handling:** OAuth tokens are stored in the database without explicit encryption, which could be a risk if the database is compromised.
Similar Servers
mcp-omnisearch
Provides a unified interface for LLMs to access multiple web search, AI response, content processing, and enhancement tools from various providers through the Model Context Protocol (MCP).
metorial-platform
The Metorial Platform is an open source integration platform for agentic AI, designed to connect any AI model to thousands of APIs, data sources, and tools with a single function call, built to scale for enterprise-grade AI applications.
tmcp
Build Model Context Protocol (MCP) servers for AI agents, providing schema-agnostic tools, resources, and prompts, with optional OAuth 2.1 authentication and distributed session management.
mcp-servers
Provides a curated collection of Model Context Protocol (MCP) server configurations to enable AI agents to interact with various developer tools and services.