Skip to main content
On Silvana, agents can apply defined logic to asset operations, coordinate settlement, and automate multi-step financial processes while remaining within configured execution rules. This guide shows how to take an agent from initial setup to a working Silvana workflow, whether you start from an existing implementation or adapt one to your own use case.

​
Path 1: Launch from Agent Space

Start by opening Agent Space, the catalog of agents available for use with Silvana.
  1. Choose an agent from the catalog.
  2. Open its details and select an available template.
  3. Test the demo, when provided.
  4. Copy the implementation, adjust its settings, and run it on your infrastructure.
If an existing implementation matches your workflow, copy its code and adapt the configuration. Otherwise, build a custom agent with the SDK.

​
Path 2: Build on Silvana Book Agent

Use cloud-agent to configure or extend the existing Silvana Book Agent for your own trading and settlement workflows.

​
1. Install Silvana Book Agent

For workflows based on Silvana Book Agent, download the latest cloud-agent binary from GitHub Releases. Choose the archive for your platform. Checksums are provided on the release page. Alternatively, build the CLI from source:
The compiled binary is available at:
Use the Silvana Book Agent Quickstart for platform-specific installation commands.

​
2. Onboard the Agent

Each new agent must complete onboarding before it can submit authenticated operations. Run:
The onboarding flow registers the agent and creates the Canton party used for subsequent operations.
Silvana Book Agent connects to the following RPC endpoint by default:
Use the —rpc option to specify another endpoint when required.

​
3. Configure the Agent

Define the agent’s operating rules, including its markets, strategy parameters, execution cadence, limits, counterparties, RPC endpoint, and failure-handling behavior.
Silvana Book Agent uses agent.toml for its runtime configuration. Custom implementations can extend the repository or use the SDK and proto definitions directly.

​
4. Test the Workflow

Test the decision logic independently before allowing the agent to submit transactions.
Verify that the agent:
  1. connects to the configured endpoint;
  2. completes onboarding successfully;
  3. retrieves balances and required service information;
  4. follows the configured markets, strategy parameters, and limits;
  5. processes order and settlement updates correctly;
  6. handles rejected or failed operations safely.

​
5. Run the Agent

After onboarding, configuration, and testing, run the agent on your infrastructure. For the long-running Silvana Book Agent mode:
Production deployments should use secure key storage, logging, health monitoring, and a defined shutdown or kill-switch process. Configure only the operations required for the agent’s workflow.
For direct programmatic access to Silvana Ledger through gRPC, see the Agentic API documentation.

Silvana Book Agent Quickstart

Set up and launch the agent

Configuration

Adjust runtime settings

SDK Intro

Explore the SDK

Getting Started Agentic API

Connect to the Agentic API