Skip to main content
The Grounded Docs MCP Server provides a powerful CLI for managing documentation indexes, running servers, and querying documentation.

​
Installation

Install globally via npm: 
npm install -g docs-mcp-server
Or use via npx: 
npx docs-mcp-server <command>

​
Quick Start

# Scrape and index documentation
docs-mcp-server scrape react https://react.dev/reference/react

# Search the indexed documentation
docs-mcp-server search react "hooks"

# List all indexed libraries
docs-mcp-server list

# Start the MCP server (default command)
docs-mcp-server

​
Available Commands

CommandDescription
server (default)Start the unified MCP server
scrapeScrape and index documentation from a URL or local directory
searchQuery the documentation index
listList all indexed libraries and versions
removeDelete a library’s documentation from the index
refreshUpdate an existing library by re-scraping changed pages
configView or modify configuration
mcpStart MCP server in standalone mode
webStart web interface only
workerStart pipeline worker in standalone mode
fetch-urlFetch and display a single URL (debugging)
find-versionFind the best matching version for a library

​
Global Options

These options are available for all commands:
​
--verbose
boolean
default:"false"
Enable verbose (debug) logging
​
--silent
boolean
default:"false"
Disable all logging except errors (mutually exclusive with --verbose)
​
--telemetry
boolean
Enable/disable telemetry collection. Use --no-telemetry to disable.
​
--store-path
string
Custom path for data storage directory. Defaults to system-specific location. Can also be set via DOCS_MCP_STORE_PATH environment variable.
​
--config
string
Path to configuration file. If specified, the config is read-only and cannot be modified via config set.
Show ASCII art logo on startup (for server commands)
​
--help
boolean
Show help information for a command
​
--version
boolean
Show version information

​
Common Patterns

​
Scraping Local Documentation

Use the file:// protocol to scrape local HTML files or directories: 
# Single HTML file
docs-mcp-server scrape mylib file:///Users/me/docs/index.html

# Directory of HTML files
docs-mcp-server scrape mylib file:///Users/me/docs/my-library
When using Docker, mount the local directory and use the container path.

​
Working with Versions

Version matching supports flexible patterns: 
# Exact version
docs-mcp-server search react "hooks" --version 18.0.0 --exact-match

# Semantic version range (matches 18.0.0 or earlier)
docs-mcp-server search react "hooks" --version 18.0.0

# Wildcard patterns
docs-mcp-server search typescript "types" --version 5.x
docs-mcp-server search typescript "types" --version 5.2.x

​
Using External Pipeline Workers

For distributed setups, connect CLI commands to a remote pipeline worker: 
# Scrape using remote worker
docs-mcp-server scrape react https://react.dev \
  --server-url http://localhost:8080/api

# Search using remote worker
docs-mcp-server search react "hooks" \
  --server-url http://localhost:8080/api

​
Custom Embedding Models

Specify embedding models per command: 
# Use OpenAI embeddings
docs-mcp-server scrape react https://react.dev \
  --embedding-model openai:text-embedding-3-small

# Use Vertex AI embeddings
docs-mcp-server scrape react https://react.dev \
  --embedding-model vertex:text-embedding-004

​
Output Formats

The config command supports multiple output formats: 
# JSON output (default)
docs-mcp-server config get scraper.maxPages --json

# YAML output
docs-mcp-server config --yaml

# Plain text for scalars
docs-mcp-server config get scraper.maxPages

​
Error Handling

​
Common Issues

Mutually Exclusive Options 
# Error: Cannot use both --verbose and --silent
docs-mcp-server --verbose --silent list
Invalid Configuration Path 
# Error: Use a more specific path
docs-mcp-server config set scraper "value"

# Correct: Set a leaf value
docs-mcp-server config set scraper.maxPages 200
Read-Only Configuration 
# Error: Cannot modify when using explicit --config
docs-mcp-server --config /path/to/config.yaml config set scraper.maxPages 200

# Correct: Remove --config to modify default config
docs-mcp-server config set scraper.maxPages 200

​
Exit Codes

  • 0 - Success
  • 1 - Error (invalid arguments, operation failed, etc.)

​
Shell Completion

The CLI is built with Yargs, which supports shell completion. Generate completion scripts: 
# Bash
docs-mcp-server completion >> ~/.bashrc

# Zsh
docs-mcp-server completion >> ~/.zshrc

​
Next Steps

Scrape Command

Index documentation from URLs or local files

Search Command

Query your documentation index

Configuration

Manage server configuration

Environment Variables

Configure via environment variables

Build docs developers (and LLMs) love

Get started for freeTalk to us