# Genaro

> Genaro is managed content and generative AI infrastructure for software built by developers and coding agents. It provides workspace-isolated media storage, image and video generation, durable operations, signed webhook events, usage attribution, and prepaid billing through one HTTP API.

Genaro customers build their own products and workflows. They do not run Genaro code, share their downstream user identities, or manage GPU infrastructure. Use the OpenAPI document as the canonical interface contract.

## Documentation

- [Documentation](https://genaro.ai/docs): Quickstart, core concepts, integration workflow, reliability guidance, and API endpoint reference.
- [Complete agent reference](https://genaro.ai/llms-full.txt): Self-contained implementation context for coding agents.

## Machine-readable contracts

- [OpenAPI 3.1](https://genaro.ai/openapi.yaml): Canonical request, response, parameter, and schema definitions.
- [LLM documentation map](https://genaro.ai/llms.txt): This concise documentation index.

## Console

- [Create a developer account](https://genaro.ai/sign-up): Create an infrastructure account and workspace API key.
- [Developer sign in](https://genaro.ai/sign-in): Access workspaces, balance, API keys, models, and recent generations.

## Key integration facts

- API base URL: `https://api.genaro.ai/api/v1`
- Authentication: `Authorization: Bearer <workspace-api-key>`
- Executable models (generated from the live catalog; `GET /models` returns the same list): `krea-2`, `minimax-h3`, `seedream-v5-pro`, `kling-video-v3-0-pro`, `topaz-video-upscale`, `nano-banana-pro`, `kling-image-v3`, `z-image-turbo`, `topaz-image-upscale`
- `availability: preview` means the model runs and is priced but has no quality or latency guarantee yet; every listed model is `executable: true`. `GET /models?include_hidden=true` also returns catalog entries that cannot run (`executable: false`), which are rejected with `422 model_unavailable`, and models held back from the public list.
- Generation requests are asynchronous. Each `POST /generations` creates a new generation.
- Register a webhook endpoint and select the generation event types your application needs.
- Read the same durable lifecycle records through the Event API.
- Attach independent captions, scores, labels, regions, and notes to Assets with Annotations.
- Organize Assets into typed, ordered, editable Groups without implying lineage.
- Register an OTLP/HTTP endpoint to receive terminal Generation traces.
- Verify webhook timestamps and HMAC signatures against the exact raw request body, and deduplicate on `webhook-id`: failed deliveries are retried for about 24 hours and can be listed and replayed per endpoint.
- Every API key belongs to exactly one isolated workspace.
- Use `metadata` for customer-defined identifiers and context; Genaro returns it unchanged.
- Failed generations receive a full customer refund.
- Customer-requested cancellation is not supported in v1.