SDK Guide
Install the tooling, onboard an agent, configure a strategy and run the first working flow.
Best for: getting an agent running fast
Developers
From strategy to running agent.
Build agents and applications on Silvana with a typed gRPC API, an agent focused SDK and deployment models that fit your control boundary.
Prototype locally. Sign in your environment. Deploy on your terms.
Start here
Choose the path that matches how you build.
API Reference
Explore gRPC services, transaction flow, streaming models and structured responses.
Best for: direct integration
Playground
Configure a strategy and watch it react against a simulated market before writing code.
Best for: understanding behavior first
Deploy Agents
Choose where the agent runs and where data, secrets, signing and operations belong.
Best for: production planning
Build flow
One path from first config to production.
The developer experience should feel direct. Start with a working pattern, inspect the behavior, configure the mandate, connect services and deploy the same logic in the right runtime.
Prototype
Start with an included agent flow, example configuration or direct API integration.
Simulate
Use the Playground or mock data to inspect strategy behavior before live execution.
Configure
Set markets, permissions, limits, strategy parameters and signing behavior.
Connect
Use SDK or gRPC clients to reach orderbook, settlement, ledger and event services.
Authorize
Prepare and verify sensitive transactions before signing inside your environment.
Deploy
Run the agent in the infrastructure model that fits your operating requirements.
Controlled by design
Every state change follows the same flow.
Silvana is built for agents that can act continuously without receiving blind authority over assets.
Prepare
Silvana builds the transaction and returns the payload, hash, identifiers and estimated requirements.
Verify
Your application checks the operation, recipient, asset, amount and expected outcome.
Sign
The approved environment signs the prepared transaction. Private key material stays inside that environment.
Execute
The signed transaction is submitted through Silvana and routed to the relevant Canton workflow.
Reconcile
Your system receives status, update references, contracts, usage information and errors.
What you build with
The pieces an agent needs to operate.
SDK and CLI
Onboard agents, manage configuration and start from ready trading and settlement workflows.
Agentic API
Prepare and execute payments, transfers, DvP, multicall and application operations.
Proto and gRPC
Generate typed clients and connect from any backend stack with gRPC support.
Orderbook services
Access market data, submit orders, manage RFQs and subscribe to execution events.
Settlement streams
Follow proposals, allocations, lifecycle events and final settlement status.
Verification and signing
Inspect prepared transactions and authorize them within the agent or client environment.
Run on your terms
Same agent logic. Different control boundaries.
When the agent is ready to leave local development, choose the runtime model that matches your security, confidentiality and operating requirements.
Self hosted
Run the agent inside infrastructure your organization owns and operates.
Cloud deployment
Deploy in your cloud account or approved cloud project.
Silvana managed
Use a managed environment when speed and lower operational overhead matter.
Trusted Execution Environment
Run supported logic and sensitive material inside an isolated runtime when confidentiality requirements are higher.
Go deeper
Use the reference that matches the task.
Documentation
Read setup, authentication, transaction flow, signing and settlement guides.
Quickstart
Move from installation to a running agent through a guided first implementation.
API Reference
Inspect services, operations, streaming models and response fields.
GitHub
Explore examples, releases and developer tooling.
Start building
Build the first version today.
Start with the SDK guide, connect to the API, test behavior in the Playground and choose the runtime when the agent is ready to operate.