• Old host target: http://localhost:4020
• New host target: http://localhost:4024
• Auth token source for authenticated checks: POST /api/v1/auth/dev/impersonate on the existing API at http://localhost:4001

Parity claimed here is parity for current search, suggest, list, and auth-protected admin behavior. It is not a claim that migration ownership cleanup or final legacy-host retirement is already done.

Routes Checked

GET /butkhuzi/norms?page=1&pageSize=1

• Old host result: 200, {"ok":true,"norms":[],"meta":{"total":0,"page":1,"pageSize":1,"totalPages":0,"count":0,"filters":{}}}
• New host result: 200, same body
• Parity verdict: confirmed

GET /butkhuzi/suggest?query=SNIP&limit=2

• Old host result: 200, {"ok":true,"suggestions":[],"meta":{"query":"SNIP","count":0,"limit":2,"sourceKey":null,"sourceVersion":null}}
• New host result: 200, same body
• Parity verdict: confirmed

GET /butkhuzi/search?query=SNIP&page=1&pageSize=2

• Old host result: 200, {"ok":true,"chunks":[],"meta":{"total":0,"page":1,"pageSize":2,"totalPages":0,"count":0,"query":"SNIP","sourceKey":null,"sourceVersion":null,"langCode":null,"chunkType":null}}
• New host result: 200, same body
• Parity verdict: confirmed

POST /butkhuzi/upsert without bearer token

• Request body used: {}
• Old host result: 401, {"ok":false,"reasonCode":"missing_bearer_token"}
• New host result: 401, same body
• Parity verdict: confirmed

POST /butkhuzi/chunks/rebuild with bearer token

• Request body used: {"sourceKey":"SNIP"}
• Old host result: 200, {"ok":true,"deleted":0,"inserted":0,"meta":{"sourceKey":"SNIP","sourceVersion":null}}
• New host result: 200, same body
• Parity verdict: confirmed

Mismatches Found

• None in the checked set.

Notes

• The checked dataset on this local environment returned empty search, suggest, and norms results on both hosts. That still counts as parity because status, shape, and payload matched.
• The authenticated admin-path check succeeded on both hosts using the same dev impersonation token flow.
• This sprint verified coexistence parity only. It did not prove live cutover, startup-flow normalization, or legacy-host retirement.