Skip to main content

MCP Commands

The mcp command group manages Model Context Protocol servers, which extend your LLM's capabilities with external tools.

Overview

lc mcp <subcommand> [options]

Subcommands

add - Add an MCP Server

Add a new MCP server to your configuration.

lc mcp add <name> <command> --type <type> [-e KEY=VALUE]...

Arguments:

  • <name> - Unique identifier for the server
  • <command> - Command to start the server or URL for remote servers
  • --type - Server type: stdio, sse, or streamable

Options:

  • -e, --env KEY=VALUE - Set environment variables (can be used multiple times)

Examples:

# Add fetch server for internet access
lc mcp add fetch "uvx mcp-server-fetch" --type stdio

# Add Playwright server for browser automation
lc mcp add playwright "npx @playwright/mcp@latest" --type stdio

# Add remote SSE server
lc mcp add remote "http://localhost:8080/sse" --type sse

# Add server with environment variables
lc mcp add kagi "uvx kagimcp" --type stdio -e KAGI_API_KEY=your_api_key

# Add server with multiple environment variables
lc mcp add myserver "uvx myserver" --type stdio -e API_KEY=key -e DEBUG=true

delete (alias: d) - Delete an MCP Server

Remove an MCP server from your configuration.

lc mcp delete <name>

Example:

lc mcp delete fetch

list (alias: l) - List MCP Servers

Display all configured MCP servers.

lc mcp list

Example output:

Servers: Configured MCP servers:
• fetch - Stdio (uvx mcp-server-fetch)
• playwright - Stdio (npx @playwright/mcp@latest)

functions (alias: f) - List Server Functions

Show available functions/tools provided by an MCP server.

lc mcp functions <name>

Example:

lc mcp functions fetch
# Output:
# Functions: Available functions:
# • fetch - Fetches a URL from the internet and optionally extracts its contents as markdown.
# Parameters: max_length, raw, start_index, url

invoke (alias: i) - Invoke a Function

Call a specific function from an MCP server directly.

lc mcp invoke <server> <function> [parameters...]

Parameters format: key=value

Examples:

# Fetch a URL
lc mcp invoke fetch fetch url=http://example.com

# With multiple parameters
lc mcp invoke fetch fetch url=http://example.com max_length=1000

# Browser automation
lc mcp invoke playwright navigate url=https://google.com

stop (alias: st) - Stop an MCP Server

Close a persistent MCP server connection managed by the daemon.

lc mcp stop <name>

Example:

lc mcp stop fetch

Note: With the new SDK implementation, MCP servers are started automatically when needed. The stop command is only needed if you want to explicitly close a persistent connection.

Using MCP Tools in Prompts

Once MCP servers are configured, use them in your prompts with the -t or --tools flag:

Direct Prompts

# Single tool
lc -t fetch "What's the weather in Paris?"

# Multiple tools
lc -t fetch,playwright "Go to weather.com and get the forecast"

Interactive Chat

# Start chat with tools
lc chat -m gpt-4 -t fetch

# Multiple tools in chat
lc chat -m claude-3-opus -t fetch,playwright

Common MCP Servers

mcp-server-fetch

Provides internet access to fetch web content.

# Install
lc mcp add fetch "uvx mcp-server-fetch" --type stdio

# Test
lc mcp invoke fetch fetch url=http://httpbin.org/json

# Use in prompt
lc -t fetch "Summarize the latest tech news"

Playwright MCP

Browser automation and web scraping.

# Install
lc mcp add playwright "npx @playwright/mcp@latest" --type stdio

# Test
lc mcp invoke playwright screenshot url=https://example.com

# Use in prompt
lc -t playwright "Take a screenshot of google.com"

Context7 MCP

Access library documentation and code examples.

# Install
lc mcp add context7 "npx -y @upstash/context7-mcp" --type stdio

# Test
lc mcp functions context7

# Use in prompt
lc -t context7 "Show me React useState examples"
lc -t context7 "How to use Express.js routing"

File System MCP

Read and write local files.

# Install
lc mcp add fs "uvx mcp-server-fs" --type stdio

# Use in prompt
lc -t fs "Read the contents of config.json"

Kagi MCP

Access Kagi search and summarization capabilities.

# Install (requires Kagi API key)
lc mcp add kagi "uvx kagimcp" --type stdio -e KAGI_API_KEY=your_kagi_api_key

# Test
lc mcp functions kagi

# Use in prompt
lc -t kagi "Search for the latest AI developments"
lc -t kagi "Summarize this article: https://example.com/article"

Note: Requires Python 3.12+ and a valid Kagi API key from https://kagi.com/settings?p=api

Configuration

MCP configurations are stored in:

  • macOS/Linux: ~/.config/lc/mcp.toml or ~/Library/Application Support/lc/mcp.toml
  • Windows: %APPDATA%\lc\mcp.toml

Example configuration:

[servers.fetch]
name = "fetch"
server_type = "Stdio"
command_or_url = "uvx mcp-server-fetch"

[servers.fetch.env]

[servers.playwright]
name = "playwright"
server_type = "Stdio"
command_or_url = "npx @playwright/mcp@latest"

[servers.playwright.env]

[servers.kagi]
name = "kagi"
server_type = "Stdio"
command_or_url = "uvx kagimcp"

[servers.kagi.env]
KAGI_API_KEY = "your_api_key_here"

Troubleshooting

Server Won't Start

  • Ensure the command is installed (uvx, npx, etc.)
  • Check if the command works when run directly
  • Check environment variables are set correctly if required
  • The server starts automatically when you use it with -t flag or invoke a function

Connection Issues

  • For HTTPS issues, try HTTP URLs when possible
  • Check firewall/proxy settings
  • Verify network connectivity

Function Invocation Fails

  • Check parameter names and types with lc mcp functions <name>
  • Ensure boolean parameters are not quoted
  • The server will be started automatically when needed

Examples

Web Research

# Add fetch server
lc mcp add fetch "uvx mcp-server-fetch" --type stdio

# Research a topic
lc -t fetch "Research the latest developments in quantum computing"

Automated Testing

# Add browser automation
lc mcp add playwright "npx @playwright/mcp@latest" --type stdio

# Test a website
lc -t playwright "Navigate to my website at example.com and check if the login button is visible"

Multi-Tool Workflows

# Use multiple tools together
lc -t fetch,fs "Fetch the latest documentation from https://docs.example.com and save it to docs.md"

See Also