Three ways to build on your AI visibility data

Yes. The AI search visibility API returns JSON over plain REST, so anything that can call an HTTP endpoint can read it: Looker Studio, Power BI, Tableau, Metabase, or a warehouse job on a schedule. Pull visibility scores, mention rates, ranking positions, and cited sources per brand and per engine, then join them against your own tables. Scoped read-only keys (`read:brands`, `read:reports`) mean a BI connector never gets write access.

The full reference lives at /docs/v1/, generated from an OpenAPI 3.1 specification. Every endpoint documents its parameters, response schema, and error codes, with runnable curl, JavaScript (fetch), and Python (httpx) examples. Because the spec is machine-readable, you can also generate a typed client in any language with openapi-generator, Stainless, Speakeasy, or Fern.

Yes. One request returns AI visibility data across all six engines: ChatGPT, Claude, Gemini, Perplexity, Google AI Mode (AI Overviews), and Grok. They are normalized into a single schema, so you compare positions and citations across engines without writing per-platform parsers or maintaining six different scrapers.

Yes. Competitive AI search visibility data is part of the report payload: alongside your own brand, reports return the competing brands each engine named for the same prompt, their position in the answer, and the sources cited for them. That lets you track share of voice over time and alert when a competitor overtakes you in a category.

Close to it. Reports run on demand or on a schedule, and outbound webhooks push `report.completed`, `report.failed`, `schedule.run.completed`, and `schedule.run.failed` to your endpoint the moment a run finishes, with HMAC-signed payloads. You do not need to poll: your systems learn about new AI visibility data as soon as it exists.

Yes. Per-brand key restriction lets you mint a key scoped to a single brand, which is what makes white-label and per-client enterprise dashboards practical: each customer sees only their own AI visibility data. Combined with scoped permissions, idempotent writes, and a stable error envelope, the API is built to sit behind custom integrations you support in production.

Brand management (create, list, retrieve, update, delete), AI visibility reports across the six major LLM platforms, citation & source tracking with the Rank Prompt Score, scheduled reports, outbound webhooks, plus the account endpoints (`/v1/me`, `/v1/me/quota`). You can also connect an MCP client. Brand audits, AI-optimized content generation, technical SEO audits, and outreach are on the public roadmap and ship under the same /v1 contract; subscribe to the changelog to track release dates. The eventual surface mirrors the dashboard 1:1.

Every plan includes Developer API access, free included, so you can mint a key straight from the Developers page in your workspace. Want a larger dedicated quota? Subscribe to a standalone API Starter or API Pro plan. Each key carries scoped permissions (e.g. `read:brands`, `write:brands`) and an optional brand restriction for white-label use. Keys are prefixed with `rp_live_`, are shown once at creation, and stored hashed in the database.

Developer API access is included with every plan, with a monthly request cap per tier: 1,000 on Free, 10,000 on Starter, 25,000 on Pro, 50,000 on Agency, and 100,000 on Agency Plus. Credit-consuming /v1 endpoints (the ones that hit a paid LLM or third-party search backend) draw from your credit balance, so you never pay twice; on Free there are no monthly credits, so you top up before running billable work. Want a bigger dedicated quota? Two standalone subscriptions cover that: API Starter ($99/mo, 50,000 request units, 500 bundled API credits) and API Pro ($299/mo, 200,000 request units, 2,000 bundled API credits). Every plan applies a 120 requests/minute per-key burst limit, and the per-endpoint cost map lives in the developer docs.

No. Developer API access is included with your plan, with a monthly request cap that scales by tier (1,000 on Free, then 10,000 / 25,000 / 50,000 / 100,000). Credit-consuming /v1 endpoints draw from the same monthly credits you already pay for, so there's nothing extra to buy. You'd only add a standalone API Starter or API Pro plan if you want a larger dedicated request quota with its own bundled API credits.

A per-key burst limit of 120 requests per rolling minute is applied to every API key, regardless of plan. On top of that, your plan grants a monthly request-unit quota. The quota belongs to the account: every key made in it, yours and those your team made, draws on one pool, and the account owner sees and can revoke any of them. Every metered response includes `X-RP-Quota-Limit`, `X-RP-Quota-Used`, `X-RP-Quota-Remaining`, and `X-RP-Quota-Reset` so you can pace work without polling. Throttled calls return a 429 with `Retry-After`.

Not yet. The API ships with an OpenAPI 3.1 spec, so you can generate idiomatic clients in any language using openapi-generator, oazapfts, hey-api, Stainless, Speakeasy, or Fern. The docs include curl, JavaScript (fetch), and Python (httpx) examples for every endpoint. First-party TypeScript and Python SDKs are on the roadmap.

The `GET /v1/me` and `GET /v1/me/quota` endpoints cost 0 request units and never count against your quota, so you can validate keys, scopes, and request signing without burning budget. For full end-to-end tests, mint a separate key with limited scopes and exercise it against a test brand.

The /v1 contract is additive. We only add fields and endpoints; we never silently remove or rename them. When a breaking change is unavoidable, we ship /v2 alongside /v1 with a long deprecation window, and we announce the sunset of any endpoint via the standard `Sunset` HTTP header well in advance.

Yes. Optionally bind a key to a single brand at creation time (`allowed_brand_id`). The API will refuse any request whose target brand does not match, even if the calling key carries `write:all`. This lets agencies and platforms issue per-brand keys to clients without spinning up a new Rank Prompt workspace.

Yes, webhooks are live. Register a per-brand endpoint with `POST /v1/brands/{brand_id}/webhooks` and subscribe to `report.completed`, `report.failed`, `schedule.run.completed`, and `schedule.run.failed`. Payloads are HMAC-signed, the signing secret rotates (the old secret stays valid for 24 hours), and an endpoint auto-disables after repeated failures. You no longer need to poll report status from a worker.

Yes. The Rank Prompt MCP server lets Claude Code, Cursor, Claude Desktop, or any MCP client read your reports, brands, and citations, and run reports on your behalf. You connect over OAuth in the browser, so there is no key to paste. See the MCP server page at rankprompt.com/mcp for the 30-second setup.

Bearer-token authentication over TLS, scoped per-resource. Browser CORS calls are blocked at the edge: the API is server-to-server only. Keys are stored hashed and we submit the `rp_live_` prefix to GitHub secret scanning so leaked keys are caught automatically. The API runs on the same hardened infrastructure as the rest of the Rank Prompt platform, with EU/US data residency options.