Skip to content

README.md

Latest commit

 

History

History
247 lines (172 loc) · 5.92 KB

README.md

File metadata and controls

247 lines (172 loc) · 5.92 KB

AgentCore Gateway Integration

Add Valyu search tools to your AWS Bedrock AgentCore Gateway.

Why Use Gateway?

  • Centralized API Keys - Store Valyu API key in gateway config, not in application code
  • Audit Logging - Track all tool calls through AWS CloudTrail
  • Unified Access - Single gateway endpoint for multiple MCP tool providers
  • Enterprise Auth - Use your existing Cognito/OAuth infrastructure

Quick Start

Add to Existing Gateway (Recommended)

If you already have an AgentCore Gateway:

from valyu_agentcore.gateway import add_valyu_target

add_valyu_target(gateway_id="your-gateway-id")

Done. Valyu tools are now available to any agent connected to your gateway.

Create New Gateway

For testing or new deployments:

from valyu_agentcore.gateway import setup_valyu_gateway, GatewayAgent

# One-time setup (creates gateway + Cognito auth)
setup_valyu_gateway()

# Use the agent
with GatewayAgent.from_config() as agent:
    response = agent("Search for NVIDIA SEC filings")
    print(response)

Installation

pip install valyu-agentcore[agentcore]

Set your Valyu API key:

export VALYU_API_KEY=your-api-key

Get your key at platform.valyu.ai.

Available Functions

add_valyu_target

Add Valyu to an existing gateway:

from valyu_agentcore.gateway import add_valyu_target

result = add_valyu_target(
    gateway_id="your-gateway-id",
    valyu_api_key="your-key",  # or uses VALYU_API_KEY env var
    region="us-east-1",
    target_name="valyu-search",
)

print(f"Target ID: {result['target_id']}")

setup_valyu_gateway

Create a new gateway with Cognito authentication:

from valyu_agentcore.gateway import setup_valyu_gateway

config = setup_valyu_gateway(
    valyu_api_key="your-key",  # or uses VALYU_API_KEY env var
    gateway_name="my-gateway",
    region="us-east-1",
)

print(f"Gateway URL: {config.gateway_url}")
# Config saved to valyu_gateway_config.json

GatewayAgent

Use after setup_valyu_gateway():

from valyu_agentcore.gateway import GatewayAgent

with GatewayAgent.from_config() as agent:
    # List available tools
    tools = agent.list_tools()
    print(f"Available: {len(tools)} tools")

    # Run a query
    response = agent("Research Tesla's financial performance")
    print(response)

cleanup_valyu_gateway

Remove gateway resources:

from valyu_agentcore.gateway import cleanup_valyu_gateway

cleanup_valyu_gateway()  # Uses valyu_gateway_config.json

Available Tools

Once Valyu is added to your gateway, these tools become available:

Tool Description
valyu_search Web search returning full page content
valyu_academic_search Academic papers from arXiv, PubMed, journals
valyu_financial_search Stock prices, earnings, market data
valyu_sec_search SEC filings (10-K, 10-Q, 8-K) with section search
valyu_patents Patent search and analysis
valyu_contents Extract content from any URL

Tools appear with a prefix based on your target name:

valyu-search___valyu_search
valyu-search___valyu_sec_search
valyu-search___valyu_financial_search

Manual Setup via AWS Console

If you prefer the AWS Console:

  1. Go to Amazon Bedrock → AgentCore → Gateways
  2. Select your gateway (or create one)
  3. Click Add target
  4. Configure:
    • Name: valyu-search
    • Type: MCP server
    • Endpoint: https://mcp.valyu.ai/mcp?valyuApiKey=YOUR_VALYU_API_KEY
  5. Wait for status to show "Ready"

Using with Your Own Agent

After adding Valyu to your gateway, use the tools from your existing agent code:

from strands import Agent
from strands.models import BedrockModel
from strands.tools.mcp.mcp_client import MCPClient
from mcp.client.streamable_http import streamablehttp_client

# Connect to your gateway
def create_transport():
    return streamablehttp_client(
        "https://your-gateway.gateway.bedrock-agentcore.us-east-1.amazonaws.com/mcp",
        headers={"Authorization": f"Bearer {your_access_token}"}
    )

with MCPClient(create_transport) as mcp:
    tools = mcp.list_tools_sync()

    agent = Agent(
        model=BedrockModel(model_id="us.anthropic.claude-sonnet-4-20250514-v1:0"),
        tools=tools,
    )

    response = agent("Find NVIDIA's latest 10-K and summarize risk factors")
    print(response)

Example Scripts

Script Description
add_valyu_target.py CLI to add Valyu to existing gateway
setup_gateway.py Create new gateway with Cognito auth
use_gateway.py Demo using GatewayAgent

add_valyu_target.py

# Add to existing gateway
python add_valyu_target.py --gateway-id your-gateway-id

# List targets
python add_valyu_target.py --gateway-id your-gateway-id --list

# Remove target
python add_valyu_target.py --gateway-id your-gateway-id --remove TARGET_ID

setup_gateway.py

# Create new gateway
python setup_gateway.py

# Cleanup
python setup_gateway.py cleanup

use_gateway.py

# Run demo
python use_gateway.py

# Interactive mode
python use_gateway.py -i

Troubleshooting

"Target synchronization failed"

  • Verify your Valyu API key is valid at platform.valyu.ai
  • Check network connectivity from AWS to mcp.valyu.ai

"Access denied" errors

  • Ensure your IAM role has bedrock-agentcore:* permissions
  • Check that your OAuth token is valid and not expired

"Cognito credentials not found"

  • Run setup_valyu_gateway() first to create the gateway
  • Check that valyu_gateway_config.json exists

Tools not appearing

  • Wait 1-2 minutes for target sync
  • Check target status in AWS Console or via --list flag

Links