MCP Integration

Overview

The Model Context Protocol (MCP) is the universal standard for connecting AI models to external tools. Jarvis SDK exposes a fully compliant MCP server that lets any MCP client discover and execute tools from our catalog.

Quickstart (2 minutes)

curl -X POST https://jarvissdk.com/api/mcp \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}}}'

curl -X POST https://jarvissdk.com/api/mcp \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'

curl -X POST https://jarvissdk.com/api/mcp \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc":"2.0",
    "id":3,
    "method":"tools/call",
    "params":{
      "name":"jarvis_text-toolkit_word_count",
      "arguments":{"text":"Hello world from Jarvis SDK"}
    }
  }'

Setup Guides

Add Jarvis SDK to your MCP client. Pick your tool below.

{
  "mcpServers": {
    "jarvis-sdk": {
      "type": "http",
      "url": "https://jarvissdk.com/api/mcp",
      "headers": {
        "x-api-key": "jsk_your_api_key_here"
      }
    }
  }
}

{
  "mcpServers": {
    "jarvis-sdk": {
      "url": "https://jarvissdk.com/api/mcp",
      "headers": {
        "x-api-key": "jsk_your_api_key_here"
      }
    }
  }
}

{
  "mcpServers": {
    "jarvis-sdk": {
      "serverUrl": "https://jarvissdk.com/api/mcp",
      "headers": {
        "x-api-key": "jsk_your_api_key_here"
      }
    }
  }
}

import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js";

const transport = new StreamableHTTPClientTransport(
  new URL("https://jarvissdk.com/api/mcp"),
  {
    requestInit: {
      headers: { "x-api-key": "jsk_your_api_key_here" },
    },
  }
);

const client = new Client({ name: "my-agent", version: "1.0.0" });
await client.connect(transport);

// List all tools
const { tools } = await client.listTools();
console.log(`${tools.length} tools available`);

// Call a tool
const result = await client.callTool({
  name: "jarvis_hash-toolkit_sha256",
  arguments: { text: "hello world" },
});
console.log(result);

import httpx

MCP_URL = "https://jarvissdk.com/api/mcp"
HEADERS = {
    "x-api-key": "jsk_your_api_key_here",
    "Content-Type": "application/json",
}

def mcp_call(method: str, params: dict = None, req_id: int = 1) -> dict:
    payload = {"jsonrpc": "2.0", "id": req_id, "method": method}
    if params:
        payload["params"] = params
    r = httpx.post(MCP_URL, json=payload, headers=HEADERS)
    return r.json()

# Initialize
init = mcp_call("initialize", {
    "protocolVersion": "2024-11-05",
    "capabilities": {},
    "clientInfo": {"name": "my-python-agent", "version": "1.0"},
})

# List tools
tools = mcp_call("tools/list", req_id=2)
print(f"{len(tools['result']['tools'])} tools available")

# Execute a tool
result = mcp_call("tools/call", {
    "name": "jarvis_text-toolkit_slugify",
    "arguments": {"text": "Hello World from Python"},
}, req_id=3)
print(result["result"])

Available Tools

When you call tools/list, Jarvis SDK returns every module action and published skill as individual tools. Tool names follow these conventions:

Popular tool categories

MCP Resources

Jarvis SDK also exposes resources/list and resources/read for reading module metadata, schema definitions, and trust scores without executing anything.

Working Examples

Hash a string

// tools/call
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "jarvis_hash-toolkit_sha256",
    "arguments": { "text": "hello world" }
  }
}

// Response
{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "content": [{
      "type": "text",
      "text": "{\"hash\":\"b94d27b9934d3e08a52e52d7da7dabfac484efe37a5380ee9088f7ace2efcde9\"}"
    }]
  }
}

Execute a composed skill

// tools/call — skills run multiple modules in sequence
{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "jarvis_skill_content-publisher",
    "arguments": {
      "topic": "AI agent marketplaces",
      "tone": "professional"
    }
  }
}

// Response includes step-by-step results
{
  "jsonrpc": "2.0",
  "id": 2,
  "result": {
    "content": [{
      "type": "text",
      "text": "{\"skill\":\"Content Publisher\",\"status\":\"completed\",\"steps_completed\":\"3/3\", ...}"
    }]
  }
}

Agent conversation flow

1. jarvis_json-toolkit_validate({ json: configFileContents })
   → { valid: true, parsed: { ... } }

2. jarvis_hash-toolkit_sha256({ text: configFileContents })
   → { hash: "a1b2c3..." }

Error Handling

Jarvis SDK returns standard JSON-RPC 2.0 error codes. Your MCP client library handles these automatically, but here's what they mean:

Tool execution errors return isError: true in the tools/call result rather than a JSON-RPC error, following MCP convention. The error message is in the content array.

Rate Limits

MCP requests count against your plan's rate limits. Each tools/call counts as one execution.

tools/list responses are cached for 60 seconds server-side, so frequent polling is efficient. initialize and metadata methods do not count against execution limits.

Troubleshooting

Supported JSON-RPC Methods