Interoperability

PsyCloud's authoring surfaces — Studio, the Python SDK, the R SDK, and imports from other frameworks — all converge on one bundle format. That shared format is what makes round-tripping possible.

The common artifact

Whatever you author with, the result is a bundle: a set of JSON artifacts (bundle.json, designProgram.json, screenProgram.json, bindings.json, and optionally counterbalancePolicy.json) plus any assets. The runtime, Studio, and the CLI all read this same format, so a study's source (Studio vs. code) doesn't change how it runs.

Code → Studio

A bundle authored with psycloudpy or psycloudr can be opened in Studio:

1. Write the bundle

   write_bundle(...) (R) or write_bundle(...) / pc_bundle_write(...) (Python) emits the bundle directory.

2. Import it

   In Studio, choose New Study → Import bundle (.zip from psycloudr / psycloudpy), or use the Import page. Studio reads the design program and bindings and builds an editable study.

3. Keep editing visually

   From there it's a normal Studio study — preview, launch, and monitor as usual.

Studio → Code

You can also go the other direction. The CLI's export-code command generates idiomatic SDK code from an imported study or bundle, in any of the three languages:

# Generate Python (psycloudpy)
psycloud export-code ./study.json --lang python -o design.py
 
# Generate R (psycloudr)
psycloud export-code ./study.json --lang r -o design.R
 
# Or all surfaces at once
psycloud export-code ./study --lang all -o ./export/

The output uses the fluent style of each SDK (Python method chaining, R |> pipelines, or the @psycloud/authoring-ts contract helpers), so it's a real starting point you can keep editing in code.

Validate anywhere with the CLI

Because the format is shared, the same CLI commands work on any bundle regardless of how it was authored:

psycloud validate ./my-study      # structural validation
psycloud inspect  ./my-study      # show the design/screens/bindings
psycloud compile  ./my-study      # compile to a runnable Study Bundle

This is what lets you put authoring in CI: generate a bundle from code, validate and compile it, and fail the build on any regression.

Next