raindrop mcp

The raindrop mcp commands integrate Raindrop with AI coding assistants through Model Context Protocol (MCP). Install Raindrop support for Claude Code to build and manage Raindrop applications with AI assistance.

MCP integration provides:

Context-aware assistance - AI assistants understand your Raindrop project structure
Slash commands/recipes - Quick workflows for common tasks
Session management - Resume previous development sessions
Manifest generation - AI-assisted raindrop.manifest creation

Supported AI Assistants

Assistant	Install Command	Requirements
Claude Code	`raindrop mcp install-claude`	Claude Code

Prerequisites

Before installing MCP integration:

Authenticate with Raindrop:
Terminal window
```
raindrop auth login
```
Install target AI assistant (Claude Code)
Verify installation:
Terminal window
```
# For Claude
claude --version
```

Command Reference

raindrop mcp status

Check MCP integration status for all supported AI assistants.

Syntax:

raindrop mcp status [--output <format>]

Flags:

Flag	Description	Values	Default
`-o, --output`	Output format	`text`, `json`	`text`

Examples:

# Check status
raindrop mcp status

# Get status as JSON
raindrop mcp status --output json

Output:

Raindrop MCP Integration Status

Claude Code:
  Status: Installed
  MCP Server: raindrop-mcp
  Slash Commands: 3 installed
    - /new-raindrop-app
    - /update-raindrop-app
    - /reattach-raindrop-session

raindrop mcp install-claude

Install complete Raindrop integration for Claude Code IDE.

Syntax:

raindrop mcp install-claude

Flags: None (uses default configuration)

Examples:

raindrop mcp install-claude

What Gets Installed:

MCP Server Configuration
- Adds raindrop-mcp server to Claude Code’s MCP configuration
- Configures authentication using your Raindrop credentials
- Location: ~/.claude/mcp_servers.json
Slash Commands (in ~/.claude/commands/):
- /new-raindrop-app - Start new Raindrop application development
- /update-raindrop-app - Update existing Raindrop applications
- /reattach-raindrop-session - Resume previous development sessions

Output:

Installing Raindrop integration for Claude Code...

✓ Checking Claude Code installation
✓ Verifying Raindrop authentication
✓ Adding raindrop-mcp server to MCP configuration
✓ Creating slash commands:
  - /new-raindrop-app
  - /update-raindrop-app
  - /reattach-raindrop-session

Installation complete!

Try it out:
1. Open Claude Code
2. Type /new-raindrop-app to start a new project
3. Follow the AI-guided workflow

Using MCP Integration

Starting a New Project

With Claude Code:

1. Open Claude Code
2. Type: /new-raindrop-app
3. Follow the prompts:
   - Application name
   - Template selection
   - Initial configuration

Claude will:
- Generate raindrop.manifest
- Create project structure
- Set up handlers
- Initialize git repository

Updating an Existing Application

With Claude Code:

1. Open project directory in Claude Code
2. Type: /update-raindrop-app
3. Describe the changes you want:
   "Add a new HTTP handler for user authentication"
   "Configure a PostgreSQL database"
   "Add environment variables for API keys"

Claude will:
- Analyze current manifest
- Suggest changes
- Update manifest
- Generate/update code
- Run build/deploy if requested

Resuming a Session

All assistants support session resumption for continuing previous work:

Claude Code:

/reattach-raindrop-session

Claude will:
- Load previous session context
- Show last actions taken
- Continue where you left off

MCP Capabilities

When MCP integration is installed, AI assistants can:

Project Understanding

Read and parse raindrop.manifest
Understand application structure
Identify handlers, databases, services
Map dependencies

Code Generation

Generate handler implementations
Create type definitions
Write database schemas
Build configuration files

Manifest Management

Create new manifests from scratch
Validate manifest syntax
Suggest configuration improvements
Add new resources

Build and Deploy

Run raindrop build generate
Execute raindrop build deploy
Monitor deployment status
Handle errors and retry

Context Preservation

Remember previous commands
Track changes made
Maintain session state
Resume interrupted work

Troubleshooting

Installation Issues

“Claude Code not found”:

# Install Claude Code
# Download from: https://claude.com/download

# Verify installation
claude --version

“Not authenticated with Raindrop”:

# Log in first
raindrop auth login

# Verify authentication
raindrop auth list

“MCP server already exists”:

The installation detected an existing raindrop-mcp server.
This prevents duplicate installations.

To reinstall:
1. Remove existing configuration manually
2. Run install command again

Runtime Issues

Slash commands not appearing:

# Restart your AI assistant
# For Claude Code:
# Close and reopen the application

# Verify installation
raindrop mcp status

“Permission denied” errors:

# Check file permissions
ls -la ~/.claude/commands/

# Fix permissions if needed
chmod 755 ~/.claude/commands/*.md

MCP server not responding:

# Check Raindrop authentication
raindrop auth list

# Re-authenticate if needed
raindrop auth logout
raindrop auth login

# Reinstall MCP integration
raindrop mcp install-claude

Uninstallation

To remove MCP integration:

Claude Code:

# Remove MCP server from configuration
# Remove raindrop-mcp entry

# Remove slash commands
rm ~/.claude/commands/new-raindrop-app.md
rm ~/.claude/commands/update-raindrop-app.md
rm ~/.claude/commands/reattach-raindrop-session.md

Best Practices

Development Workflow:

Start new projects with /new-raindrop-app for proper setup
Use /update-raindrop-app for iterative changes
Leverage session resumption for long-running tasks
Let AI handle boilerplate code generation

Collaboration:

Share AI-generated manifests with team
Review AI suggestions before applying
Use version control for all AI-generated code
Document AI-assisted changes in commits

Security:

Never commit credentials or secrets
Review AI-generated environment variable configurations
Validate MCP server connections
Keep AI assistant software updated

Exit Codes

Code	Description
0	Command completed successfully
1	Installation failed (AI assistant not found, not authenticated, etc.)
2	Invalid arguments