Unreleased documentation. These pages follow the main branch and can change before the next release. For supported guidance, use v0.15.2.
Start from a working local sample, then replace its synthetic workbook with one reviewed non-personal workbook. Registry Stack does not infer access policy or safe fields from source data. You make those decisions in small reviewed project files.
Start from the protected sample
Section titled “Start from the protected sample”Complete Run your first registry API before adapting a source. The sample gives you a known-good project, a loopback-only runtime, and maintained denied and allowed checks.
Use a spreadsheet
Section titled “Use a spreadsheet”Choose this path when the source is one reviewed .xlsx workbook with a
stable key, one selected worksheet, and a fixed set of columns.
Use your own spreadsheet walks from the working sample to a strictly described workbook and protected API.
Explore advanced source patterns
Section titled “Explore advanced source patterns”Registry Stack also documents bounded HTTP, reviewed script, exact snapshot, and Fast Healthcare Interoperability Resources (FHIR) source patterns. No end-to-end tutorial is published for these advanced options.
Use integration patterns to compare their boundaries. Use the configuration reference to inspect the exact authored fields and validation metadata.
Know what you own
Section titled “Know what you own”| Surface | What it controls | Who changes it |
|---|---|---|
| Workbook copy | The source rows and columns used by this project | Source owner |
| Entity file | The stable key, fields, types, and row limits | Project reviewer |
| Project file | The service purpose, caller scopes, filters, and returned fields | Project reviewer |
| Environment file | The workbook location, worksheet, callers, and local runtime | Environment operator |
| Generated build and runtime directories | Product configuration and local runtime state | registryctl; do not edit |
Keep real credentials and personal data out of tutorials and fixtures. Use a reviewed non-personal derivative when you test institution-owned data.