generated: '2026-09-05'
method: searched
source: https://github.com/7Signal/API-Examples/blob/develop/docs/02-rate-limiting.md
docs: https://github.com/7Signal/API-Examples/blob/develop/docs/02-rate-limiting.md
limit_count: 1
algorithm: token bucket
enforcement: automatic, no opt-in or configuration
rate_limits:
- scope: per-api-key
  algorithm: token bucket
  window: continuous replenishment (per second)
  limit: null
  burst: null
  limit_note: >-
    7SIGNAL documents the MECHANISM and the runtime headers but does not publish the numeric bucket
    size or replenish rate — those are set per account type/plan and are only observable from the
    response headers. The worked example in the docs shows burst-capacity 15 and replenish-rate 5/s,
    but the docs present it as an illustration, not a published entitlement, so it is not recorded as
    the limit. "Rate limits can vary depending on your account type" and "Not by default" is the
    published answer on raising them.
  varies_by:
  - account type
  - plan
  - api key
  status_on_exhaustion: 429
  headers:
  - name: ratelimit-remaining
    meaning: tokens left in the bucket before the limit is hit
  - name: ratelimit-burst-capacity
    meaning: maximum tokens that can be accumulated at once
  - name: ratelimit-replenish-rate
    meaning: tokens added back per second
  - name: ratelimit-requested-tokens
    meaning: tokens consumed by this specific request
  header_note: >-
    The reference chapter documents these header names unprefixed (`ratelimit-*`); the repository's own
    DEVELOPER_GUIDE.md and examples/rate_limiting/rate_limit.py read them with an `x-` prefix
    (`x-ratelimit-remaining`, etc.). The two 7SIGNAL sources disagree, so a client should read both
    spellings. Neither is RFC 9238/draft-ietf-httpapi-ratelimit-headers shaped (no `RateLimit-Limit`,
    `RateLimit-Reset` or `Retry-After` is documented).
  retry_after: not documented
  guidance:
  - Wait a few seconds before retrying, then back off further if the 429 persists
  - Watch ratelimit-remaining to monitor headroom
  - Cache responses and avoid unnecessary calls
  notification_before_limit: none — the only signal is the ratelimit-remaining header
  reference_implementation: https://github.com/7Signal/API-Examples/blob/develop/examples/rate_limiting/rate_limit.py
observed:
  probed: false
  note: >-
    Rate-limit headers were not observed on a live response because every gateway path except
    /api/gateway-v2.json and the static /swagger-ui/ tree returns 401 without a token.