Anonymize JSON for Safe Use in Development Environments

A dev seed where orders reference users by email. hash keeps the join intact after scrubbing.

Input:
{ "users":  [ { "id": 1, "email": "ada@x.com" } ],
  "orders": [ { "id": 9, "userEmail": "ada@x.com" } ] }

Strategy: hash · Deep: on

Output:
{ "users":  [ { "id": 1, "email": "e91a77c2" } ],
  "orders": [ { "id": 9, "userEmail": "e91a77c2" } ] }

Add password and apiToken to the term list and use remove so they never reach git.

PII terms: email, name, password, apiToken
Strategy: remove

Input:
{ "name": "Ada", "password": "hunter2", "apiToken": "sk_live_…", "role": "admin" }

Output:
{ "role": "admin" }

When you need a record that still LOOKS like a real user in the dev UI, mask keeps length and format.

Strategy: mask

Input:
{ "fullName": "Grace Hopper", "phone": "+1 212 555 0143" }

Output:
{ "fullName": "Gr*********er", "phone": "+1 212 555 *143" }  (all but last 4 digits masked)

Substring matching is broad. If your dump has username/filename keys you want to keep, use narrower terms.

PII terms: name
Strategy: hash

Input:
{ "name": "Ada", "username": "ada99", "filename": "avatar.png" }

Output (all three matched 'name'):
{ "name": "…", "username": "…", "filename": "…" }

Output is 2-space pretty-printed. Minify it for a compact seed asset using a sibling tool.

Anonymizer output (pretty):
{
  "users": [ { "email": "e91a77c2" } ]
}

→ run json-minifier →
{"users":[{"email":"e91a77c2"}]}

Free tier caps the file at 2 MB. A full production dump usually exceeds this — slice it to a sample or upgrade to Pro (100 MB) / Developer (5 GB).

Some hand-edited fixtures contain trailing commas or comments. JSON.parse rejects them and nothing runs. Clean it first with json-format-fixer.

fake increments a counter, so the same source email becomes user1@…, user2@… — the join breaks. For committed seeds with relations, use hash, not fake.

With Deep off, only top-level keys are evaluated; a profile.email two levels down is copied untouched. Keep Deep on for ORM dumps that nest relations.

If your dev code requires the key (NOT NULL column, required prop), remove will produce records that fail to load. Use hash or fake to keep the key present instead.

Detection never reads values, so an email inside a description string stays. Add description to the list and use remove/fake, or filter it with json-key-filter.

Removed keys are not counted in 'fields anonymized'. The count reflects retained anonymized values only — a low number after a remove run is normal.

Values are stringified before hashing/masking, so a numeric userId becomes a string token. If your seed loader is strict about types, keep ids out of the PII list or cast on load.

An empty PII list disables the Anonymize button. Provide at least one term.