mojawave-mcp

MojaWave MCP server — connect any MCP-compatible AI assistant to the MojaWave SMS and Email API.

Works with Claude (Desktop & Code), ChatGPT (via OpenAI Agents SDK), Gemini (via Google ADK), Cursor, Windsurf, and any other tool that speaks the Model Context Protocol.

Every tool maps to a documented endpoint of the MojaWave public API — nothing undocumented is exposed.

Available tools

SMS

Email

Messages & account

Inputs are validated before any request is made (E.164 phone numbers, 1–11-char sender IDs, message length, recipient count, ISO-8601 schedule times, email addresses, subject length), and the client retries 429/5xx responses with backoff that honours Retry-After.

Related MCP server: Twilio SMS Server

Recommended workflows

Sending SMS

1. list_sms_sender_ids()
   → [{ "sender_id": "MYAPP", "status": "approved" }]

2. send_sms(to="+255712345678", message="Hello!", sender_id="MYAPP")
   → { "id": "...", "status": "sent" }

Sending email

1. list_email_domains()    — confirm your domain is "verified"
2. list_email_senders()    — pick a valid from_email address
3. send_email(to="customer@example.com", from_email="noreply@yourdomain.com",
              subject="Hello", body="Hi there")
   → { "id": "...", "status": "queued" }

Bulk SMS

1. list_sms_sender_ids()   — pick an approved sender ID

2. send_bulk_sms(recipients=["+255700000001", ...], message="...", sender_id="MYAPP")
   → { "job_id": "ec0fb57c-...", "status": "scheduled", "total_recipients": 500 }

3. get_bulk_sms_job(job_id="ec0fb57c-...")
   → { "status": "completed", "total_recipients": 500, "total_credits_cost": 500 }

Installation

pip install mojawave-mcp

Or for local development:

git clone https://github.com/mojawave/mojawave-mcp
cd mojawave-mcp
pip install -e ".[dev]"

Configuration

Copy .env.example to .env and add your API key:

cp .env.example .env

MOJAWAVE_API_KEY=sk_live_mw_xxxxxxxxxxxxxxxxxxxx

Get your API key from the MojaWave dashboard under Settings → API Keys.

Use a test key (sk_test_mw_…) during development — it returns synthetic responses without sending real messages or charging credits.

Connecting to AI assistants

Claude Desktop

Add this block to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "mojawave": {
      "command": "mojawave-mcp",
      "env": {
        "MOJAWAVE_API_KEY": "sk_live_mw_xxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

Restart Claude Desktop. You will see a MojaWave tool icon in the chat interface.

Claude Code (CLI)

claude mcp add mojawave -- env MOJAWAVE_API_KEY=sk_live_mw_xxx mojawave-mcp

Cursor / Windsurf / any stdio MCP client

Point the client at the mojawave-mcp command with your API key as an environment variable. Most clients use the same JSON config format as Claude Desktop above — refer to your client's MCP documentation.

OpenAI Agents SDK (ChatGPT / GPT-4o)

Start the server in SSE mode so OpenAI can reach it over HTTP:

MOJAWAVE_API_KEY=sk_live_mw_xxx mojawave-mcp --transport sse --port 8080

Then connect from Python:

from agents import Agent, Runner
from agents.mcp import MCPServerSse

async def main():
    server = MCPServerSse(url="http://localhost:8080/sse")
    async with server:
        agent = Agent(
            name="MojaWave Agent",
            model="gpt-4o",
            mcp_servers=[server],
        )
        result = await Runner.run(
            agent, "Send an SMS to +255712345678 saying Hello from AI"
        )
        print(result.final_output)

Google Gemini (Google ADK)

Start the server in SSE mode:

MOJAWAVE_API_KEY=sk_live_mw_xxx mojawave-mcp --transport sse --port 8080

Then connect from Python:

from google.adk.agents import LlmAgent
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset, SseServerParams

mojawave_tools = MCPToolset(
    connection_params=SseServerParams(url="http://localhost:8080/sse")
)

agent = LlmAgent(
    model="gemini-2.0-flash",
    name="mojawave_agent",
    instruction="You can send SMS and email and check credits via MojaWave.",
    tools=[mojawave_tools],
)

Hosted deployment (Docker)

For production, run the SSE server behind a reverse proxy:

FROM python:3.12-slim
RUN pip install mojawave-mcp
ENV MOJAWAVE_API_KEY=""
EXPOSE 8080
CMD ["mojawave-mcp", "--transport", "sse", "--port", "8080"]

docker build -t mojawave-mcp .
docker run -e MOJAWAVE_API_KEY=sk_live_mw_xxx -p 8080:8080 mojawave-mcp

Running locally (stdio)

MOJAWAVE_API_KEY=sk_live_mw_xxx mojawave-mcp

The server reads JSON-RPC from stdin and writes to stdout — the standard MCP stdio transport used by Claude Desktop and most IDE extensions.

Security notes

• Never commit your API key. Use environment variables or a secrets manager.

• Use test keys (sk_test_mw_…) in CI/CD and development — no real messages are sent and no credits are charged.

• Scope API keys to only the permissions they need from the MojaWave dashboard.

• The AI is instructed to always confirm recipient, content, and sender with you before calling send_sms, send_bulk_sms, or send_email — these spend real credits and deliver real messages.

• Webhook payloads are signed with X-MojaWave-Signature (HMAC-SHA256) — verify signatures on your server before trusting delivery events.

License

MIT