nexus-agent-sdk 1.0.0

Official Python SDK for NEXUS AI Agent Marketplace - Discover, test, and integrate APIs for AI agents

NEXUS AI Agent Marketplace - Python SDK

The official Python SDK for integrating AI agents with the NEXUS marketplace. Discover, test, and use 29+ premium APIs with simple pay-per-request pricing.

Features

• 🔍 Service Discovery - Search and filter 29+ production-ready APIs
• 🆓 Free Sandbox - 50 free calls per tool to test before you pay
• 💰 Pay-per-Request - Only pay for what you use with x402 payments
• 🔌 MCP Compatible - Works with any MCP-compatible agent
• ⚡ Async Support - Both sync and async clients available
• 🔗 Framework Integration - LangChain, LlamaIndex, AutoGPT ready

Installation

pip install nexus-agent-sdk

For async support:

pip install nexus-agent-sdk[async]

Quick Start

from nexus_sdk import NexusClient

# Initialize client
client = NexusClient()

# Discover available tools
tools = client.list_tools()
print(f"Found {len(tools['tools'])} tools")

# Try a tool for free (sandbox mode)
result = client.execute_sandbox(
    tool_name="weather-forecast-api",
    parameters={"location": "San Francisco"},
    agent_id="my-agent-001"
)
print(result)

Available APIs

Category	Examples	Price Range
AI/ML	Sentiment, Translation, NER, Summarization	0.4 - 2.0 credits
Finance	Crypto prices, Stock data, Forex rates	0.1 - 0.5 credits
Data	News, Company info, WHOIS lookup	0.2 - 1.5 credits
Utility	Weather, Geocoding, Email validation	0.1 - 0.8 credits
Communication	SMS, Email sending	0.5 - 2.0 credits

Usage Examples

Search for Services

# Search by keyword
results = client.search(query="weather")

# Filter by category
ai_services = client.search(category="ai")

# Get sandbox-enabled services only
free_to_try = client.search(sandbox=True)

# Combined filters
results = client.search(
    protocol="mcp",
    category="finance",
    max_price=1.0,
    limit=10
)

Use Sandbox (Free Tier)

Every tool offers 50 free sandbox calls:

# No payment required!
result = client.execute_sandbox(
    tool_name="sentiment-analysis",
    parameters={"text": "I love this product!"},
    agent_id="my-agent"
)

print(result)
# {'sentiment': 'positive', 'score': 0.95, 'remaining_calls': 49}

Make Paid Calls

# Generate payment proof
proof = client.generate_payment_proof(
    agent_id="my-agent",
    amount=0.5
)

# Execute with payment
result = client.execute_tool(
    tool_name="weather-forecast-api",
    parameters={"location": "Tokyo"},
    agent_id="my-agent",
    payment_proof=proof['proof']
)

Async Client

import asyncio
from nexus_sdk import AsyncNexusClient

async def main():
    async with AsyncNexusClient() as client:
        # All methods are async
        tools = await client.list_tools()
        result = await client.execute_sandbox(
            tool_name="crypto-price-tracker",
            parameters={"symbol": "BTC"},
            agent_id="async-agent"
        )
        print(result)

asyncio.run(main())

Framework Integrations

LangChain

from langchain.tools import Tool
from nexus_sdk import NexusClient

client = NexusClient()

# Create LangChain tools from NEXUS services
tools = []
for service in client.list_tools()['tools']:
    tool = Tool(
        name=service['name'],
        description=service['description'],
        func=lambda params, sid=service['service_id']: 
            client.execute_sandbox(sid, params, "langchain-agent")
    )
    tools.append(tool)

LlamaIndex

from llama_index.tools import FunctionTool
from nexus_sdk import NexusClient

client = NexusClient()

tools = []
for service in client.list_tools()['tools']:
    tool = FunctionTool.from_defaults(
        fn=lambda **kwargs: client.execute_sandbox(
            service['service_id'], kwargs, "llamaindex-agent"
        ),
        name=service['name'],
        description=service['description']
    )
    tools.append(tool)

API Reference

Client Methods

Method	Description
list_tools()	List all available MCP tools
search(**filters)	Search services with filters
execute_sandbox(tool, params, agent_id)	Free sandbox call
execute_tool(tool, params, agent_id, proof)	Paid tool execution
trending(limit)	Get trending services
get_service(id)	Get service details
get_service_metrics(id)	Get usage metrics

Error Handling

from nexus_sdk import NexusClient, NexusError, NexusRateLimitError

client = NexusClient()

try:
    result = client.execute_sandbox(...)
except NexusRateLimitError as e:
    print(f"Rate limited. Retry after {e.retry_after} seconds")
except NexusError as e:
    print(f"Error: {e.message}")

Getting Started Credits

New agents automatically receive 100 free credits on first API call, plus 50 free sandbox calls per tool. No credit card required!

Support

• 📖 Documentation
• 💬 Discord Community
• 🐛 Issue Tracker
• 📧 Email Support

License

MIT License - see LICENSE for details.