MCP Servers
The Model Context Protocol (MCP) lets Claude Code connect to external tools and data sources. On ShellPod, MCP servers run alongside your Claude Code session on the same container, giving you fast, persistent access to extended capabilities.
What is MCP?
MCP (Model Context Protocol) is an open protocol that allows AI models to interact with external systems. Think of MCP servers as plugins for Claude Code — they extend what Claude can do beyond reading files and running commands.
For example, an MCP server could let Claude Code:
- Query a documentation database for up-to-date API references
- Access your project management tool to read and update tickets
- Interact with a database to query or modify data
- Call external APIs to fetch real-time information
Available MCP servers
ShellPod comes with Context7 pre-configured as an MCP server. Context7 provides Claude Code with access to up-to-date documentation for popular libraries, frameworks, and tools. Instead of relying on training data that might be outdated, Claude Code can query Context7 for the latest docs.
This is particularly useful when working with:
- Rapidly evolving frameworks (Next.js, React, SvelteKit)
- Libraries with frequent breaking changes
- APIs whose documentation has changed since Claude's training cutoff
- Niche libraries that may not have been well-represented in training data
Using Context7
Context7 works transparently with Claude Code. When you ask Claude Code to implement something using a specific library, it can automatically query Context7 for the latest documentation. You do not need to do anything special — just ask Claude Code to build what you need.
You can also explicitly ask Claude Code to check documentation:
claude "Check the latest Next.js 15 docs for the new caching behavior and update our data fetching to match"
Claude Code will use Context7 to fetch the current documentation and apply the correct patterns.
Configuring MCP servers
MCP servers are configured in your Claude Code settings. The configuration lives in ~/.claude/settings.json on your container. You can view and edit it:
cat ~/.claude/settings.json
A typical MCP server configuration looks like:
{
"mcpServers": {
"context7": {
"command": "npx",
"args": ["-y", "@context7/mcp-server"]
}
}
}
Adding custom MCP servers
You can add additional MCP servers to your ShellPod environment. Since you have a full Linux container with Node.js and Python installed, you can run any MCP server that works on Linux.
Popular MCP servers you might want to add:
- Filesystem MCP: Give Claude Code scoped access to specific directories
- PostgreSQL MCP: Let Claude Code query and modify your database
- GitHub MCP: Create PRs, manage issues, and interact with GitHub directly
- Slack MCP: Send notifications or read messages from Claude Code
To add an MCP server, edit your settings file and add the server configuration. After saving, restart Claude Code for the changes to take effect.
MCP servers persist
Since your ShellPod container is persistent, your MCP server configurations and any locally installed MCP server packages persist across sessions. Set them up once, and they are always available.
This is a significant advantage over running Claude Code locally where you might need to restart MCP servers every time you open a new terminal, or keep them running as background processes that consume local resources.
Troubleshooting MCP servers
If an MCP server is not working:
- Check the Claude Code output for MCP connection errors when it starts up
- Verify the MCP server package is installed:
npx -y @context7/mcp-server --version - Check your settings file for JSON syntax errors
- Run
/doctorin Claude Code to check for configuration issues - Try restarting Claude Code — MCP servers connect at startup