Skip to main content
This page records changes to the Sendora v1 public API that can affect integrations. Endpoint additions are generally backward-compatible; behavior and contract changes include migration notes.

October 2026

  • Changed: credits can be decimal. Email verification costs half a credit ($0.005) for each conclusive answer, so `credits_charged` from `verify-emails` and the wallet balance can have up to four decimals (for example 8.5). A whole amount is still returned as a whole number. Sending an email costs $0.005 (1 credit for every two emails).
  • Consolidated: similar resources now share one set of calls, picked by a type field. /lead-collections covers lead lists, segments, pipeline stages, lead sources and lead requests; /work-items covers approvals, escalations, tasks and meetings (and view=attention for the needs-attention list); /connections covers sending accounts, email domains, caller IDs, sender pools, connected CRMs and integration health. Each has list or read (GET), create (POST), update (PATCH /{id}), delete (DELETE /{id}?type=) and a step call (POST /{id}/actions). Chat agents are now part of /agents (type chat on create, update, context and actions), and GET /knowledge-bases takes source_id to read one source’s editable copy and version, while PATCH /knowledge-bases/{kb_id} also restores a source’s last synced snapshot (restore_snapshots) or accepts a synced update (accept_provider_updates). Webhook endpoints and connected accounts joined /connections; team members, roles and billing reads moved into /workspace-settings; phone number search, details, settings, Twilio import and ordering documents moved onto the /phone-numbers calls; client details, limits, profile and people moved onto /partner/clients; package versions onto POST /partner/packages; pricing into PUT /partner/pricing; lead context onto /leads; voice test values onto PATCH /agents/{id}; SMS readiness onto GET /sms-registrations?view=readiness; voice cloning onto POST /custom-voices (type clone); and quotes onto POST /purchases (step quote). Campaign folders, templates, settings and import joined /campaigns (type folder, template or import; include=config.*; config on update), tools and website widgets share /agent-resources (type tool or widget), and deliverability tests joined /connections (type test, with test blacklist or placement). The documented reference went from 197 operations to 89. The earlier paths such as /segments and /accounts keep working unchanged as hidden aliases, so no integration needs to change. The Sendora MCP keeps its action names; they now call the merged routes.
  • New: phone numbers end to end. Search numbers with prices (GET /phone-numbers/available), quote and buy phone numbers, domains, mailboxes and WhatsApp accounts (POST /purchases/quotes, POST /purchases, GET /purchases), read a number with history and usage (GET /phone-numbers/{id}/details), change its agent, caller ID name, verified caller ID and sender ID (PATCH /phone-numbers/{id}/settings), and retry, test, send carrier documents or release it (POST /phone-numbers/{id}/actions). Also: team roles and permissions (GET /team/access, PATCH /team/{id}/access), voice cloning (POST /custom-voices/clone), agency client details, profile, suspend and close, client people, packages and pricing, integration health (GET /integration-health) and call recording download (GET /calls/{call_id}/recording). See Workspace Operations.
  • New: day-to-day workspace operations in the API and MCP. Domains, caller IDs and deliverability tests; campaign folders, templates, config and import; segments, pipeline stages, lead sources, lead requests and bulk lead changes; voice agent test setup (/agents/{agent_id}/context and /actions); chat agent creation and configuration (/chat-agents/{agent_id}/config, /actions, /custom-actions); knowledge base sources; custom voices (/custom-voices); approvals, escalations, tasks, GET /needs-attention, meetings and conversation actions (POST /conversations/{lead_id}/actions); sender pools (/sender-pools); connected CRMs (/crm-connections); workspace settings and the team (/workspace-settings, /team); read-only GET /billing; and agency client limits (/partner/clients/{client_id}/limits). Reads take include to fetch several parts at once. See Workspace Operations.
  • New: sending accounts and email domains in the API. GET /api/v1/public/accounts lists your mailboxes, WhatsApp, LinkedIn and own SMS accounts, or reads one with health, limits, warmup strength, incidents and active hours (id and include). PATCH /api/v1/public/accounts/{id} changes the display name, sender name, signature, active hours and daily limits, and POST /api/v1/public/accounts/{id}/actions marks a reconnect prompt as seen. GET /api/v1/public/domains lists your email domains with DNS health, and include=dns adds the live records. The Sendora MCP has matching actions. See Sending Accounts and Domains.
  • New: SMS registration (10DLC) in the API. /api/v1/public/sms-registrations lists, creates, edits and deletes your registered business (type brand) and texting campaigns (type campaign) as one resource, POST /api/v1/public/sms-registrations/{id}/actions runs the steps (submit, trust check, phone code, appeal, clone, attach numbers), and GET /api/v1/public/sms-registrations/readiness says whether the workspace can text yet and quotes the fee first. The Sendora MCP has matching actions. See SMS registration.
  • New: send a one-off email to any address with POST /api/v1/public/messages (to, subject, content). Choose delivery: "scheduled" (default, paced by the mail engine and tracked) or "instant" (sent right away). The Sendora MCP has a matching send_email tool.
  • API reference restructured. The reference now lists 61 operations in 7 sections instead of 162 in 19. Related actions are merged into one endpoint: list and get are one call (GET /leads, pass id for one), settings that belong to a thing are fields on its update call (a lead’s tags, stage and traffic light on PATCH /leads/{lead_id}; a campaign’s status and enrolled leads on PATCH /campaigns/{campaign_id}; an agent’s channels, knowledge bases, tools and publish state on PATCH /agents/{agent_id}), and enrich, verify and research are one POST /leads/enrich. New top level endpoints: POST /leads, POST /searches, GET /searches/{search_id}, POST /messages, PATCH /dnc, /calls, /widgets, /files, GET /analytics, GET /partner/settings and PUT /partner/settings. Every earlier path keeps working exactly as before as a deprecated alias that is no longer listed in the reference, so no integration needs to change. A merged update runs its steps in a fixed order; if one fails, the earlier steps stay applied and the error lists them under applied.
  • New: POST /api/v1/public/leads/verify-emails checks email deliverability for up to 100 leads (1 credit for each conclusive answer; unknown answers and failed batches are free). Send an Idempotency-Key so a retry is charged once.
  • Enrich lead: when no provider is connected the response now includes a message that points to method: web_search. A web search that takes too long answers reason: timed_out instead of hanging.
  • Campaigns: replies are one setting per campaign. Create campaign, Update campaign and the campaign responses carry reply_mode (off, draft or auto), reply_agent_id and reply_note. Turning replies on without an agent answers 422. A campaign with reply_mode off is answered by a person; before, the answer came from per-channel switches and, when none was set, from the workspace’s oldest agent. Existing campaigns were moved to the setting that matches how their replies were handled. See Campaigns.
  • New: chat agents. GET /api/v1/public/chat-agents and GET /api/v1/public/chat-agents/{agent_id} read your chat agents, POST /api/v1/public/chat-agents/{agent_id}/live turns one on or off, and POST /api/v1/public/chat-agents/{agent_id}/start has an agent send the first message to a contact on SMS or WhatsApp. The routes answer 404 until chat agents are switched on for the workspace. See Chat Agents.
  • Send message to lead: email replies take a new optional conversation_id (the conversation_id shown on a message in Get conversation thread). An email reply must include it; without it the request is refused. Other channels ignore it.
  • Partner API: signup, experience (save, publish, roll back) and pricing endpoints now work with an agency API key whose creator is an owner or admin; they previously always answered 403 for API keys. Client feature settings now change the switches that actually apply to the client; public_api, white_label and ai_copilot cannot be set per client and answer 422. List agency clients no longer returns internal notes or billing identifiers.
  • Restricted workspaces: while a workspace is restricted for trust and safety reasons, sending a message, starting a direct call and starting a phone test call through the API are refused with 403, like campaign launches.
  • Lists: knowledge bases, workspace tools, webhook endpoints and connected accounts accept page and report X-Total-Count, X-Page and X-Per-Page, so lists longer than 200 are fully reachable. Get lead activity timeline takes limit (up to 500) and before, returns X-Next-Cursor when older events remain, and answers 502 instead of a partial list when it cannot load. See Pagination.
  • Idempotency: a request that was refused with 401, 402, 403, 408, 423, 425 or 429 is no longer stored for 24 hours; retrying with the same Idempotency-Key runs it again. Responses with a request-shaped error (400, 404, 409, 410, 413, 415, 422) are still replayed.
  • Rate limits: requests in a cost-based bucket (messages, imports, webhooks, configuration and others) return X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset. Every 429 carries Retry-After and a detail message. The limits table on the Rate limits page now matches the real limits.
  • Unknown request fields: bodies that contain a field the endpoint does not define are still accepted and the field is ignored. We are measuring how often this happens before rejecting such requests with 422; check your request field names against the API reference.
  • API keys created from now on show a longer key prefix in Settings, so keys can be told apart.
  • Campaigns: launching through the API now runs the same checks as the dashboard’s pre-launch check (required step fields, a published and synced voice agent, sender readiness, retired steps, workflow loops). A campaign that fails them is refused with 422 and the same plain sentences. A campaign name must contain at least one visible character. Campaign responses include a warnings list, empty unless a step needs attention (for example an email step that must be relaunched after a live edit).
  • Campaigns: pausing, resuming and stopping check the campaign’s current state, so a repeated or conflicting call returns 409 instead of leaving the campaign half changed.
  • One id everywhere. Every id the API returns can now be sent back to any endpoint that takes that kind of object, and it is the same value you see in the app address bar. Call records, knowledge base sources, tools, webhook endpoints and do-not-contact entries now return a short public id in id, and references such as lead_id, agent_id and campaign_id in call records return public ids too. Ids you stored from earlier responses keep working as input. See Identifiers.
  • Webhook events: lead_id, campaign_id and agent_id in event payloads are now the same public ids the API returns, so an event can be matched to an API object. Ids from earlier events still work as input.
  • List voice calls: the lead_id and campaign_id filters accept the same ids List all leads and List campaigns return.
  • Copilot: GET /api/v1/public/copilot/sessions/{thread_id} and the thread_id field of Stream a Copilot reply accept the thread_id the reply stream sends. The session response now contains only conversation fields.
  • New: GET /api/v1/public/campaign-folders lists campaign folders, the source of folder_id for Create campaign.
  • List connected channel accounts now lists the accounts a voice agent can send from (ready mailboxes, WhatsApp and LinkedIn accounts, phone numbers). Each id is the account_id that Save channel assignments accepts, and that endpoint now rejects accounts that are not connected in the workspace. The channel filter value for phone numbers is phone; sms_voice is still accepted.
  • Send message to lead: connection_id is marked deprecated. It never chose the sending account (the account is picked from the lead’s existing conversation). It is still accepted and ignored, so existing requests keep working.
  • List campaign leads: rows no longer carry their own id. An enrollment is identified by the campaign and the lead_id.
  • Register webhook endpoint: description is now saved. Registering an endpoint with a description previously failed with a server error.
  • Responses no longer include internal bookkeeping fields (for example the workspace’s internal id or billing references). These were never part of the documented contract.
  • Campaign trigger webhooks (the URL on a campaign’s “Webhook received” step): the success response now returns the lead’s public id as lead_id and no longer returns workflow_id. Send that lead_id back on later calls to target the same lead; lead ids from earlier responses still work.
  • Error messages are now complete sentences that never include internal ids. Match on the HTTP status code, not on the message text.
  • Bringing numbers from your own carrier account now uses POST /api/v1/public/phone-numbers/import/search, /import/confirm and /import/outbound-profile. Request and response bodies are unchanged. The earlier paths keep working with the same behavior and now return a Deprecation: true header plus a Link header that points to the new path; move to the new paths when convenient.

September 2026

  • Expanded the public contract to cover voice-agent operations, widgets, partner workflows, imports, and campaign workflow updates.
  • Added route classification and live-contract drift checks to the documentation build.
  • Added dedicated guidance for pagination, idempotency, async jobs, resource isolation, voice-agent operations, widgets, and partner APIs.
For the exact current operation set, use the API Reference.