Automation

Choose an automation interface

Choose REST, MCP, or the Python SDK for your research workflow.

On this pageChoose an interfaceUse the same research contractStart with a small pilotExtend the workflow

Neurosnap exposes the same underlying research services through browser forms, REST, an MCP server, and a Python API client. Choose the interface that fits how you want to control the work.

Choose an interface

Interface Best fit What you manage
Browser Interactive setup and result inspection Inputs, settings, and scientific review
Pipelines Repeatable connected calculations Node connections, list sizes, file selection, and campaign scope
REST API Scripts, notebooks, and custom orchestration Authentication, multipart uploads, polling, and recovery
MCP Tool-capable assistant clients Client connection, authorization, constraints, and review of agent actions
Python SDK Python integration and local scientific analysis Package version, dependencies, and whether an operation is local or hosted

The SDK’s local structure and sequence utilities do not launch cloud jobs merely because the package is installed. Conversely, a hosted submission through any interface can consume compute credits.

Use the same research contract

Discover the current service and inspect its form before constructing a payload. Use exact field titles, supported formats, and accepted molecular representations. Separate molecular preparation from transport encoding: a correctly encoded file can still describe the wrong system.

Record input identifiers, settings, returned IDs, and native output files. Define a bounded candidate count and credit budget before orchestrating several steps.

Start with a small pilot

For scripts, follow the API quickstart. For assistant clients, configure MCP. For local analysis, start with the Python SDK.

Do not implement automatic retries for billable submissions until you can reconcile uncertain responses against job history. A network timeout after submission can occur even when the server created the job.

Extend the workflow

Use the API reference to inspect status and files, and Bulk submissions to plan repeated calculations. Use Troubleshooting to distinguish input errors from failed execution and missing authorization.

Search documentation