SDK

Turn strategy into a running agent.

Build agents that trade, quote, settle and react to orderbook events without wiring Canton infrastructure from scratch.

Silvana SDK visual

Start with the CLI. Configure in agent.toml. Move from local testing to live execution.

Agent ready CLI

Onboard, configure and run agents through a developer first command line flow.

Reusable crates

Build custom agents, tools and integrations from core SDK components.

Trading modes included

Start with RFQ, grid market making, taker loops and LP workflows.

Keys stay local

Prepared transactions are verified before signing, keeping execution controlled.

Developer flow

Configure once. Run continuously.

The SDK packages orderbook access, settlement, signing, gRPC communication and agent state into one developer stack.

01

Onboard

Register the agent and generate the local configuration required to run.

02

Configure

Set market, spread, size, levels, limits and strategy behavior in agent.toml.

03

Run

Start the agent and let it submit orders, answer RFQs and stream settlement events.

04

Extend

Use reusable crates and gRPC clients to build custom logic beyond the default modes.

Ready flows

Start with proven patterns.

RFQ buy and sell

Request quotes, accept the best offer and settle through the same atomic flow.

Grid market making

Place laddered orders around a configured mid price while agents manage levels and updates.

Taker loops

Fill a target amount across available quotes and settlement opportunities.

LP mode

Run a long lived liquidity agent that quotes, answers RFQs and tracks settlement.

Use the level of control you need.

Run the ready agent

Use the cloud agent to onboard, configure and operate supported trading flows.

Configure a strategy

Tune agent.toml for markets, spreads, levels, size and risk rules.

Build with crates

Compose custom agents with reusable SDK components and gRPC clients.

Connect directly

Use the lower level API surface for custom application and infrastructure logic.

Get started

Build your first agent.

Start with the SDK guide, test behavior in the Playground, then move the same strategy toward live execution.