Autonomous Web3 Agents Built with AWS Bedrock AgentCore and Strands SDK
These articles are AI-generated summaries. Please check the original sources for full details.
AWS re:Invent 2025 - Autonomous Web3 agents on AWS (DAT457)
In this AWS re:Invent 2025 demo, presenters built a Web3 agent using Amazon Bedrock AgentCore and the Strands Agents SDK. The system demonstrated AWS KMS integration for Ethereum wallet key management and Sepolia testnet transaction signing, alongside Docker-based deployment with session isolation.
Why This Matters
Building Web3 agents requires handling unstructured blockchain data, long-term memory, and secure agent-to-agent communication. Without session isolation, sensitive user data like transaction histories and wallet keys could be exposed. The demo highlights how AWS KMS and Bedrock AgentCore address these challenges, though production systems must balance complexity with security requirements.
Key Insights
- “Docker-based deployment with session isolation for Web3 agents” (presented at AWS re:Invent 2025)
- “Sagas over ACID for e-commerce” (not directly applicable, but illustrates distributed system patterns)
- “Strands Agents SDK used by presenters for Web3 agent development”
Working Example
# create_web3_agent_memory.py
import uv
from bedrock_agent_core import MemoryClient
uv.add("boto3", "bedrock-agent-core")
memory_id = uv.run("create_web3_agent_memory")
print(f"Memory ID: {memory_id}")
# strands_agents_streaming_memory.py
from strands import Agent
from bedrock_agent_core.memory import LongTermMemoryHooks
class Web3Agent(Agent):
def __init__(self, memory_id):
self.memory = LongTermMemoryHooks(memory_id=memory_id)
self.system_prompt = "You are a Web3 assistant with blockchain data access and memory capabilities."
def get_version(self):
return "1.0.0"
Practical Applications
- Use Case: Crypto price analysis with CoinMarketCap integration via browser tools
- Pitfall: Overlooking session isolation leading to data leaks between user sessions
References:
Continue reading
Next article
AWS Graviton5: 60% Performance Boost with 20-30% Cost Savings for EC2 Workloads
Related Content
AWS re:Invent 2025: Agents, Trust, and the Future of AI
AWS re:Invent 2025 highlighted the rise of AI agents, with over 67,000 attendees and a focus on building trust in autonomous systems.
Characterizing AWS Graviton Memory Subsystems: Graviton2 vs. Graviton4 Performance
Analysis of AWS Graviton4 reveals a 79.8% increase in L1 data architectural efficiency over Graviton2 using the Arm System Characterization Tool.
AWS re:Invent 2025: Agents Take Center Stage, But Developer Focus Remains
AWS re:Invent 2025 featured a surge in agent announcements, but a core focus on developer empowerment and platform stability emerged, underscored by the launch of Nova 2 foundation models.