# MCP Client Configuration
This guide covers configuring various MCP clients to use fedramp-docs-mcp.
## Claude Desktop
### macOS
1. Open your Claude Desktop config file:
```bash
code ~/Library/Application\ Support/Claude/claude_desktop_config.json
```
2. Add the fedramp-docs-mcp server:
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"]
}
}
}
```
3. Restart Claude Desktop
### Windows
1. Open config at `%APPDATA%\Claude\claude_desktop_config.json`
2. Add the server configuration (same JSON as above)
3. Restart Claude Desktop
### Verify Installation
In Claude Desktop, type:
> "Use the fedramp-docs MCP server to run a health check"
You should see a response showing indexed file counts and repository information.
## Cursor
[Cursor](https://www.cursor.com/) supports MCP servers via project or global configuration.
**Config file location:**
- Project: `.cursor/mcp.json` (recommended)
- Global: `~/.cursor/mcp.json`
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"],
"env": {
"FEDRAMP_DOCS_AUTO_UPDATE": "true"
}
}
}
}
```
Restart Cursor after saving. You can also configure via **Cursor Settings > MCP**.
## VS Code with Continue
1. Install the [Continue extension](https://marketplace.visualstudio.com/items?itemName=Continue.continue)
2. Open Continue config (`~/.continue/config.json`)
3. Add MCP server:
```json
{
"mcpServers": [
{
"name": "fedramp-docs",
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"]
}
]
}
```
## Zed
1. Open Zed settings (`zed: open settings`)
2. Add to the `context_servers` section:
```json
{
"context_servers": {
"fedramp-docs": {
"command": {
"path": "npx",
"args": ["-y", "fedramp-docs-mcp"]
},
"settings": {}
}
}
}
```
## LM Studio
If LM Studio supports MCP servers:
1. Navigate to the MCP/Tools configuration
2. Add server with command: `npx -y fedramp-docs-mcp`
## Kiro
[Kiro](https://kiro.dev/) is AWS's spec-driven IDE with native MCP support.
1. **Open Kiro MCP settings:**
- Global: `~/.kiro/settings/mcp.json`
- Project: `.kiro/settings/mcp.json` (takes precedence)
2. **Add the FedRAMP Docs configuration:**
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"],
"env": {
"FEDRAMP_DOCS_AUTO_UPDATE": "true"
}
}
}
}
```
3. **Save the file** - Kiro automatically loads MCP servers on config change
4. **Test it** - Ask Kiro: "List all FedRAMP FRMR documents"
## Codex (OpenAI)
[Codex](https://github.com/openai/codex) is OpenAI's open-source coding agent with MCP support via TOML configuration.
**Config file location:**
- Global: `~/.codex/config.toml`
- Project: `.codex/config.toml` (takes precedence)
```toml
[mcp_servers.fedramp-docs]
command = "npx"
args = ["-y", "fedramp-docs-mcp"]
[mcp_servers.fedramp-docs.env]
FEDRAMP_DOCS_AUTO_UPDATE = "true"
```
You can also manage MCP servers via `codex mcp`.
## Windsurf
[Windsurf](https://windsurf.com/) is an AI-powered IDE with native MCP support.
**Config file location:** `~/.codeium/windsurf/mcp_config.json`
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"],
"env": {
"FEDRAMP_DOCS_AUTO_UPDATE": "true"
}
}
}
}
```
Restart Windsurf after saving.
## VS Code + GitHub Copilot
VS Code has native MCP support through GitHub Copilot (no extensions required).
**Config file location:** `.vscode/mcp.json` (workspace-scoped)
**Note:** VS Code uses `servers` (not `mcpServers`) and requires `"type": "stdio"`.
```json
{
"servers": {
"fedramp-docs": {
"type": "stdio",
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"]
}
}
}
```
After saving, Copilot will detect the new server automatically. You can manage MCP servers from the **Command Palette** (`Ctrl+Shift+P` > "MCP: List Servers").
## Gemini CLI
[Gemini CLI](https://github.com/google-gemini/gemini-cli) is Google's command-line AI agent with MCP support.
**Config file location:**
- Global: `~/.gemini/settings.json`
- Project: `.gemini/settings.json` (takes precedence)
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"],
"env": {
"FEDRAMP_DOCS_AUTO_UPDATE": "true"
}
}
}
}
```
Restart Gemini CLI after saving.
## Generic MCP Client Configuration
For any MCP client that supports stdio transport:
| Field | Value |
|-------|-------|
| Command | `npx` |
| Arguments | `-y fedramp-docs-mcp` |
| Transport | stdio |
Or if globally installed:
| Field | Value |
|-------|-------|
| Command | `fedramp-docs-mcp` |
| Arguments | (none) |
| Transport | stdio |
## Docker-Based Configuration
For clients that support Docker:
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-v", "fedramp-cache:/root/.cache/fedramp-docs",
"ghcr.io/ethanolivertroy/fedramp-docs-mcp:latest"
]
}
}
}
```
## With Environment Variables
To customize behavior:
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "npx",
"args": ["-y", "fedramp-docs-mcp"],
"env": {
"FEDRAMP_DOCS_AUTO_UPDATE": "true",
"FEDRAMP_DOCS_UPDATE_CHECK_HOURS": "12"
}
}
}
}
```
## Using a Local Build
If you've built from source:
```json
{
"mcpServers": {
"fedramp-docs": {
"command": "node",
"args": ["/path/to/fedramp-docs-mcp/dist/index.js"]
}
}
}
```
## Troubleshooting Client Issues
### Server not appearing in client
1. Check the config file syntax (valid JSON)
2. Ensure the command path is correct
3. Restart the MCP client application
### "Command not found" errors
Ensure Node.js and npm are in your PATH:
```bash
which node
which npx
```
For Docker-based setup, ensure Docker is running:
```bash
docker info
```
### Permission denied
On Unix systems:
```bash
# Ensure npm global bin is in PATH
export PATH="$PATH:$(npm config get prefix)/bin"
```
### First run is slow
The first run clones the FedRAMP/docs repository (~50MB). Subsequent runs use the cached copy.
### Tools not working
Run a health check to verify the server is properly initialized:
1. Ask your AI assistant: "Call the health_check tool from fedramp-docs"
2. Verify the response shows indexed files > 0
See [Troubleshooting Guide](troubleshooting.md) for more solutions.
