generated: '2026-08-13'
method: searched
source: https://sendpulse.com/integrations/api + openapi/*.yml + https://api.sendpulse.com/service-directory.json
base_url: https://api.sendpulse.com
authentication:
  style: Bearer token in the Authorization header
  header: 'Authorization: Bearer <token_or_key>'
  methods:
  - type: oauth2
    flow: client_credentials
    token_endpoint: https://api.sendpulse.com/oauth/access_token
    token_ttl_seconds: 3600
    params:
    - grant_type=client_credentials
    - client_id
    - client_secret
    request_content_type: application/json
  - type: api_key
    description: Static API key generated in account settings, no expiration
    max_keys: 5
    setup_url: https://login.sendpulse.com/settings/#api
  see: authentication/sendpulse-authentication.yml
idempotency:
  supported: false
  note: No Idempotency-Key (or equivalent) request header appears in any of the 19 published OpenAPI specs, and
    the docs define no retry-safety contract. The specs' own x-ai-reasoning-instructions repeatedly tell an agent
    to check for an existing record first "to prevent duplicate entries if the API does not handle idempotency"
    — an explicit admission that POSTs are not idempotent. No Idempotency pointer is emitted for this provider.
pagination:
  consistent: false
  styles:
  - style: limit/offset
    params:
    - limit
    - offset
    used_by: bulk-email (address books, campaigns), sms, smtp
    example: https://api.sendpulse.com/addressbooks?limit=10&offset=5
  - style: size/skip
    params:
    - size
    - skip
    used_by: chatbot channel specs (whatsapp, telegram, facebook, instagram, viber-chatbot, live-chat, tiktok)
  - style: cursor
    params:
    - search_after
    used_by: crm (7 operations)
  - style: start/limit
    params:
    - start
    - limit
    used_by: a few crm and popups operations
  note: Four different pagination idioms across the surface; there is no single documented convention and no response-level
    pagination envelope.
field_expansion:
  supported: false
  note: No expand / fields / include parameter is defined anywhere in the published specs.
metadata:
  supported: true
  note: CRM contacts/deals and chatbot subscribers carry user-defined attributes ("variables"/"attributes") addressed
    by ID or name.
request_id_tracing:
  supported: false
  note: No request-id / correlation-id response header is documented and none appears in the specs.
versioning:
  style: path-prefix, per service, inconsistent
  observed:
  - 'https://api.sendpulse.com (unversioned: bulk-email, smtp, sms, verifier, viber, web-push)'
  - https://api.sendpulse.com/crm/v1
  - https://api.sendpulse.com/edu/public/v1
  - https://api.sendpulse.com/fm/public/v1
  - https://api.sendpulse.com/v2/pop-ups
  - https://api.sendpulse.com/chatbots and one path per messenger channel (unversioned)
  note: Six of the nineteen services carry no version segment at all; there is no Accept-header or query-param versioning.
  see: lifecycle/sendpulse-lifecycle.yml
error_envelope:
  format: proprietary
  see: errors/sendpulse-problem-types.yml
  shape: '{"message": "...", "error_code": <int>}'
  rfc9457: false
rate_limit_signaling:
  status: 429
  headers: []
  see: rate-limits/sendpulse-rate-limits.yml
  note: No rate-limit headers are published.
content_types:
  request: application/json
  response: application/json
  note: 620 of 621 documented response bodies are application/json; one file-manager download is application/octet-stream.
agent_affordances:
  x_ai_extensions: true
  note: 'The published specs carry SendPulse-authored vendor extensions aimed at LLM callers: x-ai-description (720),
    x-ai-role (634), x-ai-capabilities (634), x-ai-responding-instructions (630), x-ai-reasoning-instructions (624)
    and x-ai-suggestions (585). This is unusual and materially improves agent usability even though the extensions
    are non-standard.'