appId JSON tag inconsistency: camelCase vs snake_case in API response

[mashenjun]

Summary

The appId field uses camelCase in JSON tags while all other API-facing fields use snake_case (agent_id, session_id, memory_type, created_at, etc.). This creates a visible inconsistency in both request and response bodies.

Current state

Direction	Behavior
Request input	Accepts both appId (primary) and app_id (legacy fallback) via AppIDLegacy field
Response output	Always returns appId (camelCase) — no dual-serialize
Clients	None of the 4 plugins or dashboard currently read appId from responses

{
  "id": "mem_xxx",
  "content": "...",
  "appId": "app-1",          // ← camelCase, inconsistent
  "agent_id": "test-agent",   // snake_case
  "session_id": "session-1",  // snake_case
  "memory_type": "insight",   // snake_case
  "created_at": "..."         // snake_case
}

Source

• server/internal/domain/types.go:41: AppID string json:"appId,omitempty"
• server/internal/handler/memory.go:33-34: dual AppID + AppIDLegacy in request struct
• Introduced by Bosn in fa23053e (2026-06-04)

Suggested approach

Leave as-is for now. Input side already handles both formats. Output side change would be breaking. Address when designing v3 API with consistent naming.