Shopify River 4-Layer Agent Architecture
Slack as interface, River as agent surface, Aquifer as foundation and a separate harness: the four layers behind an internal agent that treats a session as a durable entity.
Shopify River Brain-Hands Decoupling Architecture
Decoupling the brain from the hands: the harness calls the LLM and emits tool intents, while the sandbox executes them, so a new agent is a profile and not a platform.
Full FlowZap Code
Copy and paste the following FlowZap code into a project in your FlowZap account to see this template diagram.
dev { # Dev - Slack
n1: circle label:"@River mention"
n2: rectangle label:"Agent request"
n12: rectangle label:"Receives response"
n1.handle(right) -> n2.handle(left)
n2.handle(bottom) -> harness.n3.handle(top) [label="Prompt"]
}
harness { # Harness - Brain
n3: rectangle label:"Loads history"
n4: rectangle label:"Calls LLM"
n5: rectangle label:"Emits tool intents"
n11: rectangle label:"Responds to dev"
n3.handle(right) -> n4.handle(left)
n4.handle(right) -> n5.handle(left)
n5.handle(bottom) -> sandbox.n6.handle(top) [label="bash, edit"]
sandbox.n10.handle(top) -> n11.handle(bottom) [label="Result"]
n11.handle(top) -> dev.n12.handle(bottom) [label="Response"]
}
sandbox { # Sandbox - Hands
n6: rectangle label:"Sets up environment"
n7: rectangle label:"Runs commands"
n8: diamond label:"OK?"
n9: rectangle label:"Raw output"
n10: rectangle label:"Returns output"
n6.handle(right) -> n7.handle(left)
n7.handle(right) -> n8.handle(left)
n8.handle(right) -> n9.handle(left) [label="Yes"]
n8.handle(bottom) -> n6.handle(top) [label="No - retry"]
n9.handle(right) -> n10.handle(left)
}
Install the MCP packagenpm install -g flowzap-mcpMCP documentation
Related templates
Shopify River Session Modes Workflow
Three session modes in one agent: interactive in Slack, automated review in the pull request, and ephemeral batch runs triggered by CI — cattle, not pets.
Multi-Platform eCommerce Integration Hub
n8n as the central integration layer: one workflow decides whether an event is a product sync or an order, routes it to the right channel and closes the loop.
Agent-to-Agent USDC Payment Workflow
Coinbase AgentKit flow: a human allocates a USDC budget, the agent authenticates its Base wallet, pays an x402 invoice and processes the supplier data, with no fiat in the path.
Agent-Mediated Fiat Payment Workflow
Stripe Agent Toolkit flow: the consumer approves an agent spend, the agent requests a scoped token, passes it to the merchant and the order completes in fiat.
Coinbase Trust Flow — On-Chain Settlement
The trust layer behind agent-to-agent payments on Coinbase: the agent initiates the spend, the transaction is broadcast, the block confirms and the payment finalizes on-chain.