Environments and data
Private environment registry
Permission to run an environment does not automatically permit publishing its datasets, examples or outputs. This SAVRN Cloud guide separates draft edits, immutable releases and sharing decisions, and explains the access controls a real private registry would need beyond the preview.
SAVRN Cloud · Coming soon · Public preview · Browser-local simulation · No live compute or payments
Inspect the local collection
Open Environments and compare existing records. Review names, descriptions, ownership context and related datasets or evaluations. Use Configure to revise the rubric and limits; the local version increments. Use public or synthetic content and retain labeled configuration exports when reviewing changes.
The local registry is stored in this browser. It does not provide access isolation among multiple users, a server-side package repository or an institutional archive. A private-looking badge demonstrates intended product language; it is not evidence that authentication or encryption protects the record.
Decide what a release means
A release should identify a specific taskset, harness, verifier and runtime. Editing a draft differs from changing a version already used in an experiment. Preserve the exact references in an evaluation receipt so another reviewer can interpret the original result.
Publication is another decision. Approval to use a dataset for one project does not automatically permit sharing its environment, examples or derived outputs.
Before connected services launch
Enforce project ownership and role-based access for metadata, package files and every result artifact. Make released versions immutable and define deprecation, withdrawal and retention behavior. Review license, consent, confidentiality and attribution before publication. Support audit trails for promotion and access changes. Test that a guessed identifier cannot expose another project's collection. Keep public discovery metadata as an explicit projection of approved fields rather than an unfiltered copy of the private registry.
Build toward the work that matters.
Tell SAVRN what your institution needs to run, who reviews the results and where its data must stay. That workload defines the next service to qualify.
Discuss an AI project