Your next tool starts with understandable data.
Build toward scientific workflows that can read precise revisions, derive results and preserve their origin.
Read independently.
Portable data should be usable with ordinary tools. Local experiments demonstrate Parquet plus metadata and manifests read without ArrayBridge or OSDU credentials.
Try the portable-map reading exercise →Make the next result traceable.
Our intended developer contract links exact inputs, transformations and outputs. Scientific meaning belongs in documented data contracts, so applications and customer code can share it.
Extend deliberately.
A connector declares its direction, supported types, precision, update mechanism and conversion losses. It checks both source access and the destination permission before transferring data.
A packaged public SDK and certified connector catalogue are still planned. We are publishing concepts and bounded examples first.
Build on a result you can explain.
Our applications and domain APIs are the place to capture scientific intent: exact inputs, a versioned operation, parameters, review and exact outputs. Customer code should be able to read that history alongside the data.