Rate Limits
Memory API limits are shared by an account's API keys across owned and shared Vaults. Related routes share the counter shown below. Dashboard sessions use separate counters, so API traffic does not consume the owner's dashboard allowance. Unauthenticated requests fall back to an IP-based bucket. IPv6 addresses share a bucket within the same /64 network.
| Method and endpoint | Limit |
|---|---|
POST /v1/memory/write | 120 requests per minute |
POST /v1/memory/search | 120 requests per minute |
GET /v1/memory/list | 120 requests per minute |
GET /v1/memory/trash | Shares the list counter |
POST /v1/memory/view | 120 requests per minute |
GET /v1/memory/versions/{id} | Shares the view counter |
GET /v1/memory/versions/{id}/{revision} | Shares the view counter |
PUT /v1/memory/update | 120 requests per minute |
POST /v1/memory/restore | Shares the update counter |
DELETE /v1/memory/delete | 120 requests per minute |
GET /v1/memory/status/{memory_id} | 120 requests per minute |
GET /v1/memory/identity | 120 requests per minute |
GET /v1/memory/export | 3 requests per hour; one export at a time per account |
Limits use moving windows and count HTTP calls. Each search page is one call and one request. On HTTP 429, wait for the returned retry delay.