If you delete this record, what actually breaks?

Salesforce's own describe() data for any object already includes exactly the flags this tool needs (cascadeDelete, restrictedDelete). The safest way to get it: use your own already-logged-in sandbox session — nothing is typed anywhere, no separate login, no third-party tool involved.

Simplest, one object at a time — while logged into your sandbox, open a new tab and go directly to:
https://YOUR-INSTANCE.my.salesforce.com/services/data/v61.0/sobjects/Account/describe/ (swap in your instance and the object you want). Your browser's existing session shows you the JSON directly — select all, copy, paste below.

For several objects at once, or your whole org — use Salesforce's own Developer Console (Setup gear icon → Developer Console), not your browser's. Open Debug → Open Execute Anonymous Window, paste the Apex below, and click Execute:

Worth knowing before you run this: a real org can have thousands of standard objects, and Apex debug logs have a size limit. This script outputs a simple line-per-relationship format rather than JSON specifically because of that — if the log does get cut off, you only lose the last incomplete line, not the whole payload (JSON needs to be structurally complete to parse at all, so any truncation breaks everything). It still defaults to custom objects only — the part your org's schema this tool genuinely can't know about otherwise, since standard objects are already covered by the built-in model. Flip includeStandardObjects to true only if you're confident the total will fit in one log.

Boolean includeStandardObjects = false; // true = fetch everything (risk: debug log size limit on large orgs)
Integer objCount = 0;

for (Schema.SObjectType targetType : Schema.getGlobalDescribe().values()) {
    Schema.DescribeSObjectResult d = targetType.getDescribe();
    if (!includeStandardObjects && !d.isCustom()) continue;
    if (!d.isAccessible()) continue;

    System.debug('SCHEMA|OBJ|' + d.getName() + '|' + d.getLabel());
    objCount++;

    for (Schema.ChildRelationship cr : d.getChildRelationships()) {
        if (cr.getChildSObject() == null) continue;
        String relType = cr.isCascadeDelete() ? 'CASCADE' : (cr.isRestrictedDelete() ? 'RESTRICTED' : 'LOOKUP');
        System.debug('SCHEMA|REL|' + cr.getChildSObject().getDescribe().getName() + '|' + d.getName() + '|' + relType);
    }
}

System.debug('SCHEMA|DONE|' + objCount + ' objects fetched');

If you do need standard objects included too and hit truncation, the practical workaround is running this in batches — e.g. add a name filter like if (!d.getName().startsWith('A')) continue; temporarily, one letter range at a time, rather than everything in one log.

Then in the log panel at the bottom, tick "Debug Only" to cut the noise, and copy every line starting with SCHEMA| (the trailing DONE line confirms how many objects came through, and whether it matches what you expected). Paste those lines below exactly as they appear — one relationship per line means a truncated log just loses its last line, not the whole thing.

Only ever run this against a sandbox, never production.

Exported the whole log as a file instead (Developer Console → File → Download Log)? Drop the raw .log file below directly — no need to copy/paste anything.

Or paste text directly instead

Save imported schema in browser local storage (so it's still here next time you load this page) Parse & merge into the model