Rate Limits

Source: en/guide/rate-limits.md · Live: https://docs.1pass.dev/en/guide/rate-limits LLM-sanitized: internal links absolutized, VitePress containers → admonitions, line numbers in the Jump-to Index reference this rendered file (1-indexed).

📍 Jump-to Index

• L24-L40: ## Per-endpoint limits
• L41-L51: ## Response when exceeded
• L52-L57: ## Client recommendations
• L58-L63: ## Implementation details

---

Rate Limits

logi applies rate limits in two layers:

1. Cloudflare edge — IP-based, blocking requests before they reach the app (active in deployed environments).
2. Rails server (rack-attack) — keyed by a specific identifier (client_id, user_id, or IP), returning a JSON 429.

The Rails layer uses Rails.cache (SolidCache in production) as its store, so it works without separate Redis infrastructure.

Per-endpoint limits

Endpoint	Limit	Key	Layer	Notes
POST /session (login)	10 / 3min	IP	Rails	Brute-force defense
POST /oauth/authorize	30 / min	IP	Rails + Cloudflare
POST /oauth/token	20 / min	client_id	Rails	Falls back to IP if client_id is missing
POST /api/v1/me/otp/*	10 / min	session/user	Rails	Defends against SMS cost spikes
POST /api/v1/me/passkeys/*	20 / min	session/user	Rails
POST /api/v1/passkeys/authentication_options	20 / min	IP	Rails	Unauthenticated limit that RPs adding their own passkey login UI need to know. Issues challenges
POST /api/v1/passkeys/authenticate	15 / min	IP	Rails	Unauthenticated limit that RPs adding their own passkey login UI need to know. Verifies assertions
POST /api/v1/passkeys/authenticate	30 / min	credential id	Rails	Per-credential backstop layered with the IP limit
POST /api/v1/devices	10 / min	IP	Rails	Device-bootstrap brute-force defense
POST /api/v1/users/:sub/identity_verified	100 / hour	client_id (Bearer hash)	Rails	Per reporter app
Other /api/*	60 / min	IP	Rails	Global fallback
Static/Health (/up, /healthz)	Unlimited	—	safelist

Response when exceeded

HTTP/2 429 Too Many Requests
Content-Type: application/json

{"error":"rate_limited"}

Some endpoints may include a Retry-After header.

Client recommendations

• Exponential backoff: on a 429, prefer the Retry-After header; if it's absent, retry with widening intervals of 1 → 2 → 4 → 8 seconds.
• You'll rarely hit these limits in normal flows — persistent 429s suggest an implementation bug (a retry loop).
• For workloads where bursts are expected, consult us in advance (per-app custom limits are planned for Phase 2).

Implementation details

The server-side implementation lives in server/config/initializers/rack_attack.rb. When a limit changes, update both that file and this document together.

In the test environment, Rails.cache is null_store, so throttling is disabled. To test the rate limiting itself, swap in a MemoryStore as in spec/requests/rate_limit_spec.rb.