AI-readable Zevari documentation snapshot

Zevari Public API

Zevari exposes an HTTPS JSON REST API described by an OpenAPI 3.1 contract. The production base URL is `https://api.zevari.ai`. The API is built from the same public method contract used by Zevari's MCP-backed execution layer. Use the MCP reference when an AI assistant needs tool behavior, approval boundaries, or workflow semantics. Use this API reference when you need direct HTTP endpoints, schemas, request examples, and response shapes for a scheduled runner, CRM integration, internal tool, or agent bridge. Transport model: Zevari's public API uses endpoint-specific HTTP methods. Read operations use `GET` with query parameters. Create, action, approval, and generation operations use `POST` with a JSON body. Updates use `PATCH` with a JSON body. Deletes use `DELETE` with a JSON body when parameters are required. To use the public API, you need a Zevari account, an active workspace subscription, and permission to the workspace and LinkedIn sender or workspace resource that backs the workflow. LinkedIn-backed actions depend on the connected sender account and Zevari's documented approval and safety behavior. Approval workflow for automated runners: create or update the workspace policy with `approvalPolicy.update` when an owner/admin wants hands-off execution. `mode: "autopilot"` with `autopilot_campaigns_enabled: true` (the legacy value `autopilot_expanded` still works and means the same) approves eligible low/medium-risk actions on its own, plus eligible campaign launches and enrollments of any size and action bundles of up to 20; the Safety Center still paces what reaches LinkedIn. It does not bypass safety, warm-up, high-risk/manual boundaries, InMail, standalone public posts, paid work, tainted data, deletes, exports, or account controls. For a confirmation-gated REST write, call `confirmations.requestAction` with the exact executable payload first. If the response includes `data.autopilot.approved: true`, immediately call the write endpoint with the returned `confirmation_id`. If Autopilot is absent, false, or a policy or duplicate-withhold denial is returned, show or store `data.approval_url`, wait for the user to approve in Zevari, then retry the write endpoint with the same `confirmation_id`. REST clients should not call MCP-only chat approval tools. Personalized campaign enrollment is the convenience exception: call `campaigns.addTargets` directly with each target's complete `personalized_steps` and a stable idempotency key. That call stages `campaign_enroll_personalized_targets` itself and either applies immediately under eligible Autopilot policy or returns the confirmation and exact retry guidance. After manual approval, repeat the exact request with `confirmation_id`. Campaign creation is a two-phase workflow. `campaigns.create` saves a draft. To launch it, stage `confirmations.requestAction` with `action_type: "campaign_activate"` and `payload: { "campaign_id": "..." }`, then call `campaigns.updateStatus` with `status: "active"` and the approved `confirmation_id`. Under Autopilot with campaign launches on (`autopilot` + `autopilot_campaigns_enabled: true`, legacy `autopilot_expanded`), eligible campaign activations of any size can come back already approved. Campaign enrollment is discoverable from runtime responses. `campaigns.create` returns `status`, effective `enrollment_mode`, `targets_editable`, `add_targets_method`, and `next_action`. Enrollment defaults to `fixed`; choose `continuous` when shared-copy additions should remain open. Active fixed and continuous campaigns both accept approval-bound per-target personalized sequences without pausing. Draft or paused additions remain unscheduled. Terminal campaigns are immutable, so create a new campaign instead. Create API keys in the Zevari app at Dashboard Settings - API Keys. API keys start with `zvr_api_` and authenticate public API requests with `Authorization: Bearer <key>`. Keys are shown once, should be stored in a secret manager or environment variable, and should be rotated or revoked when access changes. Keep keys out of prompts, client-side code, CSV files, logs, and committed files. For ordered REST call flows, read API Playbooks at `https://docs.zevari.ai/api/playbooks` before wiring automation. For manual, scheduled, and hybrid workflow design, read the Workflow Guide at `https://docs.zevari.ai/help/workflows`. For agent-facing tool semantics, read `https://docs.zevari.ai/mcp/reference`. Zevari Public API v2. Production base URL: `https://api.zevari.ai`. Campaign V2 (the lead-owns-sequence product) lives here as `campaignsV2.*` — for example `POST /v2/campaignsV2/create`. Do not use frozen `/v1/campaigns.*` for new Campaign V2 work. `/v2/campaigns/submit`, `/v2/campaigns/status`, and `/v2/campaigns/cancel` are unused stubs for classic campaign async accept-and-process. They are not Campaign V2 and return not_yet_enabled. Also on `/v2`: receipts, lead action history, Sales Navigator saved searches, and `linkedin.connect-async`. Per-person variables (#1229): Campaign V2 copy for connect, message, inmail, and email_send steps may use `{{first_name}}`, `{{last_name}}`, `{{full_name}}`, and `{{company_name}}`, optionally with a fallback such as `{{first_name|there}}`. Zevari fills each lead's values at enroll. Any other `{{…}}` token is rejected when the sequence is written. A lead with no saved value and no fallback is held back from `campaignsV2.addLeads` with reason `missing_variable_value` while the other leads enroll; comment text accepts no variables. Auth: `Authorization: Bearer <Zevari API key>`. Create keys at Dashboard Settings. For Campaign V2 enroll, read `approval` on `campaignsV2.addLeads`; Manual workspaces approve in the browser then call `campaignsV2.schedule`. scheduled is not sent.

Requested URL: https://docs.zevari.ai/api/reference

API Overview

Zevari exposes an HTTPS JSON REST API described by an OpenAPI 3.1 contract. The production base URL is https://api.zevari.ai. Public API endpoints use endpoint-specific HTTP verbs: GET for reads, PATCH for updates, DELETE for deletes, and POST for creates, approvals, generation, and other actions. The API is built from the same public method contract used by Zevari's MCP-backed execution layer, so it is appropriate for scheduled runners, CRM integrations, internal tools, and agent bridges that need direct HTTP endpoints.

Access Requirements

To use the public API, you need a Zevari account and permission to the workspace resource that backs the workflow. LinkedIn-backed actions additionally require an exact funded sender seat associated with that workspace and remain subject to Zevari's documented approval and safety behavior.

Authentication

Create API keys in the Zevari app at Dashboard Settings - API Keys. API keys start with zvr_api_ and authenticate public API requests with Authorization: Bearer <key>. Keys are shown once, should be stored in a secret manager or environment variable, and should be rotated or revoked when access changes. Keep keys out of prompts, client-side code, CSV files, logs, and committed files.

OpenAPI Source

Version: v1. Production base URL: https://api.zevari.ai. This reference is generated from Zevari's curated public API contract.

Endpoints