Skip to content

Interface reference ​

Use reference pages to look up command syntax, configuration fields, responses, and events. For a first execution, follow the quickstart; for a complete research chain, follow the research workflow.

Choose an interface ​

Interface or settingReference
Shell commands and local executionCLI
Programmatic service callsPython
Client addresses, tokens, and timeoutsClient configuration
Application, component, and Job assemblyServer configuration
HTTP discovery and JobResponseAPI overview
Task submission, state, context, and graphsTask API
Live progress and log eventsSSE events
Workspace listing and previewsWorkspace API
Machines and resource readingsMachine API
Plugin management and snapshot synchronizationPlugins and synchronization API
Built-in Agent sessionsAgent API

External agents discover service tools through MCP integration. MCP exposes public Jobs; live SSE uses the HTTP event endpoints.

Read responses at the right layer ​

HTTP transport status, JobResponse success, and Task terminal state answer different questions. A successful submit returns a TaskHandle; use its task_id and run_id to wait for that execution and inspect its final state. Discover schemas on the connected service because installed plugins and configured Jobs determine available capabilities.

For implementation contracts, see Task contracts, plugin manifests, and research artifacts under Developers. For execution semantics, see Jobs and Tasks.

Agent-native quant research.