> ## Documentation Index > Fetch the complete documentation index at: https://docs.neuraldraft.io/llms.txt > Use this file to discover all available pages before exploring further. # Generate a premium video clip > Kicks off a short-form video generation. Two tiers — pick `tier=budget` (default, Wan 2.1) for **40 credits** (`video`) or `tier=premium` (Kling v2.1 / Runway Gen4) for **300 credits** (`video_premium`). Premium requires a **Build** plan or higher; Free / Hobby projects calling with `tier=premium` get `402 plan_required`. Returns a [`Job`](#tag/Jobs) with `type: video.generate`; poll `GET /jobs/{id}` for completion. ## OpenAPI ````yaml /openapi.yaml post /videos openapi: 3.1.0 info: title: Neural Draft Project API version: 1.0.1 summary: The AI-native backend platform for Lovable / Claude / v0 / Bolt-built sites. description: | The **Neural Draft Project API** is the single REST contract every AI-built site, MCP server tool call, and SDK in our ecosystem speaks against. One project key (`ndsk_live_...`) gets you CMS, blog, social, booking and commerce primitives behind a stable v1 surface. ## Quick start ```bash curl https://api.neuraldraft.io/v1/projects/me \ -H "Authorization: Bearer ndsk_live_yourkey" ``` ## Authentication Every endpoint except `GET /health` requires a project API key passed as `Authorization: Bearer ndsk_live_...`. Keys are scoped to a single project (formerly "tenant"). Resolve the project context by calling `GET /projects/me`. Keys carry one or more **scopes** (e.g. `content:write`, `commerce:admin`). A `403 Forbidden` is returned if the key lacks the scope required by the endpoint. ## Credits Most operations cost credits. Costs are documented per endpoint in this spec and on the [pricing page](https://neuraldraft.io/pricing). When a project has no credits left, every credit-consuming operation returns `402 Payment Required` with `code: insufficient_credits`. The response includes `cost` (credits the operation needed) and `balance` (credits the project had on hand) so clients can render an exact top-up prompt. Read-only endpoints (`GET`) do not consume credits. Mutating writes — upserting a content key, creating or updating a page, registering an image (URL swap or multipart upload) — cost **1 credit** each on top of the AI-generation costs documented per endpoint. ## Rate limits Default project budget is **60 requests per minute**, burst **120**. Every response includes: | Header | Meaning | |---|---| | `X-RateLimit-Limit` | Window ceiling | | `X-RateLimit-Remaining` | Requests left in window | | `X-RateLimit-Reset` | Unix seconds when bucket refills | | `Retry-After` | Seconds to wait, only on `429` | Enterprise plans get higher budgets — contact sales. ## Async work Long-running operations (blog generation, image generation, batch translation, content plans, full website generation) return a [`Job`](#tag/Jobs) resource immediately with `202 Accepted` and `status: pending`. Two ways to track progress: 1. **Poll** `GET /jobs/{id}` until `status` is one of `completed`, `failed` or `cancelled`. 2. **Stream** `GET /jobs/{id}/stream` for a Server-Sent Events feed of `progress`, `step`, `done`, `error` events. ## Webhooks Outgoing webhooks let your project URL receive events (e.g. `order.paid`, `blog_post.published`). Manage delivery URLs at [`/webhook-endpoints`](#tag/Webhooks). Every delivery is signed with HMAC-SHA256: ``` X-Neural-Draft-Signature: t=1745000000,v1= X-Neural-Draft-Event: order.paid X-Neural-Draft-Delivery: ``` Verify the timestamp is within 5 minutes and the HMAC matches before trusting the payload. ## Idempotency Mutating endpoints accept an `Idempotency-Key` header (any unique string up to 255 chars). Retries with the same key within 24 hours return the original response without re-executing the side effect. ## Errors All errors follow [RFC 7807](https://datatracker.ietf.org/doc/html/rfc7807) `application/problem+json` shape. The `code` field is a stable machine identifier; the `detail` is a human-readable explanation. termsOfService: https://neuraldraft.io/legal/terms contact: name: Neural Draft Support email: info@neuraldraft.io url: https://neuraldraft.io/support license: name: Proprietary url: https://neuraldraft.io/legal/api-license x-logo: url: https://neuraldraft.io/assets/logo-dark.svg altText: Neural Draft servers: - url: https://api.neuraldraft.io/v1 description: Production - url: https://api.staging.neuraldraft.io/v1 description: Staging security: - apiKeyAuth: [] tags: - name: Health description: Liveness, version, and the runtime OpenAPI document. - name: Projects description: > Project (formerly "tenant") metadata, API key management, and credit usage. Every API key resolves to exactly one project. - name: Brand description: | Canonical brand context — voice, colors, fonts, audience, content tone. This is the single source of truth that the MCP server's `brand://current` resource exposes to AI codegen tools. - name: Content description: | The CMS pillar. Translation keys map `dot.notation` keys to per-language values. Use `bulk` for build-time fetches of many keys at once. - name: Components description: > Editable HTML chunks registered by AI codegen tools during a build session. A registered component becomes editable in the project's admin UI. - name: Pages description: | Multi-page authoring with per-page SEO meta (meta_title, meta_description, og_*, canonical_url). Pages live alongside the components that render them. - name: Galleries description: | Named, ordered image collections. One `slug` per gallery, an `items` array of `{url, alt}` (max 200). Updates are full-replace — fetch, mutate, send the complete new list back. Use for carousels, lookbooks, product galleries, and any other variable-length image set. - name: Images description: Brand-consistent image generation, replacement, and resolution by key. - name: Blog description: | Full blog engine — posts, categories, tags, scheduling, multi-language translation, AI authoring pipeline. - name: Social description: | Social pillar — multi-platform post generation, scheduling, publishing, OAuth account connections. - name: Booking description: | Booking pillar — bookable services, weekly availability, slot lookup, bookings (admin + public), embeddable widget. - name: Commerce description: | Commerce pillar — products, variants, categories, orders, Stripe Checkout, Stripe Connect onboarding, embeddable widgets. - name: Jobs description: | Async work tracking. Every long-running AI op (blog gen, image gen, translation, content plan, website gen) returns a Job. Poll or stream. - name: Webhooks description: | Outgoing webhooks. Subscribe a URL to events your project cares about. All deliveries signed with HMAC-SHA256. - name: Forms description: | Lead-capture surfaces — newsletter subscribe and contact-form submit. Public submit endpoints (no auth, project resolved via `X-NeuralDraft-Project-Key` or `?project_id=`); admin list/delete endpoints require an API key with `forms:read` / `forms:write` scope. No credits charged — this is storage + delivery, not AI. paths: /videos: post: tags: - Videos summary: Generate a premium video clip description: | Kicks off a short-form video generation. Two tiers — pick `tier=budget` (default, Wan 2.1) for **40 credits** (`video`) or `tier=premium` (Kling v2.1 / Runway Gen4) for **300 credits** (`video_premium`). Premium requires a **Build** plan or higher; Free / Hobby projects calling with `tier=premium` get `402 plan_required`. Returns a [`Job`](#tag/Jobs) with `type: video.generate`; poll `GET /jobs/{id}` for completion. operationId: generateVideo requestBody: required: true content: application/json: schema: type: object required: - prompt properties: prompt: type: string minLength: 3 maxLength: 2000 description: | What the video should show. Use visual / cinematic language: action verbs, camera movement, mood. aspect_ratio: type: string enum: - '1:1' - '16:9' - '9:16' - '4:3' - '3:4' - '3:2' - '2:3' default: '9:16' duration_seconds: type: integer minimum: 3 maximum: 30 default: 5 visual_style: type: string maxLength: 200 platforms: type: array items: type: string enum: - facebook - instagram - twitter - tiktok - linkedin responses: '202': description: Job queued. content: application/json: schema: type: object properties: data: $ref: '#/components/schemas/Job' '402': $ref: '#/components/responses/PaymentRequired' security: - apiKeyAuth: [] components: schemas: Job: type: object required: - id - type - status - created_at properties: id: type: string example: job_2Ngd9KqLmRpW type: $ref: '#/components/schemas/JobType' status: $ref: '#/components/schemas/JobStatus' progress: type: integer minimum: 0 maximum: 100 message: type: string steps: type: array items: type: object properties: name: type: string completed: type: boolean active: type: boolean result: type: - object - 'null' additionalProperties: true description: > Result payload when `status` is `completed`. Shape depends on `type`: - `blog_post.generate` → `{ post_id, slug, title }` - `social_post.generate` → `{ social_post_id, title, platforms }` - `image.generate` → `{ url, key, width, height }` - `translation.batch` → `{ keys_translated: int, locales: [string] }` error: type: - object - 'null' properties: code: type: string example: upstream_unavailable message: type: string created_at: type: string format: date-time updated_at: type: string format: date-time JobType: type: string enum: - blog_post.generate - social_post.generate - image.generate - translation.batch - content_plan.generate - website.generate JobStatus: type: string enum: - pending - processing - completed - failed - cancelled Error: type: object description: | RFC 7807 problem+json error. The `code` is a stable machine identifier; clients should branch on `code`, never on `title`. required: - type - title - status - code properties: type: type: string format: uri example: https://api.neuraldraft.io/errors/validation_failed title: type: string example: Validation failed status: type: integer example: 422 detail: type: string example: One or more fields failed validation. instance: type: string description: Request id (also returned as `X-Request-Id` header). example: req_2Nh4PqRsTuVw code: type: string description: | Stable machine error code. Standard values: `bad_request`, `unauthorized`, `insufficient_credits`, `forbidden`, `not_found`, `conflict`, `validation_failed`, `rate_limited`, `internal_error`, `service_unavailable`, `slot_unavailable`, `connect_not_ready`, `upstream_unavailable`, `idempotency_conflict`. example: validation_failed cost: type: integer description: | Credits the operation requires. Only present on `402 insufficient_credits` responses. example: 1 balance: type: integer description: | Credits the project had on hand at the time of the request. Only present on `402 insufficient_credits` responses. example: 0 errors: type: object description: Field-level validation errors (only on 422). additionalProperties: type: array items: type: string example: customer_email: - The customer email field must be a valid email. starts_at: - The starts at field must be a valid ISO 8601 date. responses: PaymentRequired: description: | Insufficient credits. The project does not have enough credits to cover the operation. The body includes `cost` (credits the operation needs) and `balance` (credits the project had on hand) so clients can render an exact top-up prompt. Top up at `/projects/me/usage` or wait for the next period to start. headers: X-Request-Id: schema: type: string content: application/problem+json: schema: $ref: '#/components/schemas/Error' examples: broke: value: type: https://neuraldraft.com/errors/payment-required title: Payment Required status: 402 detail: This project has no credits remaining for the current period. code: insufficient_credits cost: 1 balance: 0 instance: req_2Nh4PqRsTuVw securitySchemes: apiKeyAuth: type: http scheme: bearer bearerFormat: API Key description: | Project API key. Pass as `Authorization: Bearer ndsk_live_...`. Manage keys via [`/projects/me/api-keys`](#tag/Projects). Test-mode keys use the `ndsk_test_` prefix. ````